# AgentTrading

> AgentTrading (https://agenttrading.ai) is an AI trading assistant for research and strategy analysis. A plain-English thesis or a ticker becomes a fundamentals summary, a rule-based backtest on 20+ years of split- and dividend-adjusted daily historical data, a plain-English risk explanation (drawdown, concentration, regime dependence, sample size), and an honest verdict: HELD UP, MIXED, or UNDERPERFORMED. Educational analysis only: no trade execution, no brokerage connection, no personalized investment advice, no performance promises.

AgentTrading is thesis-first, not signal-first: the user brings the idea, the assistant extracts a testable rule and shows it before running anything, and every backtest prints its assumptions (date range, 0.1% default cost per trade, parameters, dividend treatment) on the result. Verdicts are honest by design: when a rule underperforms buy-and-hold, the verdict says so. Built for active self-directed investors, swing traders, analysts, research teams, and investment clubs. Past performance does not guarantee future results.

## Availability and pricing status (read this before quoting a price)

Plans: Starter $19/mo, Analyst $49/mo, Pro $129/mo and Enterprise custom, listed on https://agenttrading.ai/pricing. Signing up takes an email address and a confirmation code.

## Key pages
- https://agenttrading.ai/ : AI trading assistant overview with the interactive Strategy Studio
- https://agenttrading.ai/how-it-works : the 5-station bench (thesis, evidence, backtest, risk, verdict)
- https://agenttrading.ai/ai-stock-analysis : fundamentals summarized with checkable claims
- https://agenttrading.ai/ai-stock-research : thesis to evidence to verdict in one pass
- https://agenttrading.ai/backtesting-software-for-financial-advisors : backtesting software for financial advisors and RIAs. Nine advisor-facing platforms checked on August 24, 2026: only Nitrogen ($149 to $495/mo on annual terms) and Portfolio Visualizer (free to 15 assets, then $30 or $55/mo, and only the $55 Pro tier is licensed for commercial use) publish a price. Morningstar Advisor Workstation, YCharts and Addepar publish none, orion.com/pricing returns HTTP 404, and kwanti.com/pricing returned HTTP 429 behind a bot check on both August 22 and August 24. Koyfin Advisor Core $209 and Advisor Pro $299 billed annually, with no backtesting at any tier. Regulatory frame: under SEC Rule 206(4)-1 paragraph (e)(8) backtested performance IS hypothetical performance, and paragraph (d)(6) requires policies and procedures on audience relevance plus disclosure of the criteria, assumptions, risks and limitations before it may appear in an advertisement. Rule effective May 4, 2021, compliance date November 4, 2022. The SEC announced charges against nine registered investment advisers on September 11, 2023 for website hypotheticals, penalties reported at $850,000 total. Not legal advice.
- https://agenttrading.ai/blog/backtesting-software-for-schwab-users : Schwab includes thinkBack, OnDemand and thinkScript Strategies free with an account, and none of the three is a general historical strategy engine. Reproducible check, August 24, 2026: schwab.com/trading/thinkorswim returned about 6,900 words of rendered copy containing ZERO occurrences of "backtest", "thinkBack" or "thinkScript", and two of "OnDemand". Schwab does not market thinkorswim as a backtesting product even though the review industry describes it as one. This does NOT mean the tools were removed; thinkBack is still in the desktop platform.
- https://agenttrading.ai/trading-strategy-tester : plain-English backtesting software and strategy tester. The rule is restated as an explicit entry-and-exit card before anything runs, then tested on 20+ years of split- and dividend-adjusted daily data with 0.1% charged per trade, with every assumption printed and an honest verdict of HELD UP, MIXED or UNDERPERFORMED stamped against buy and hold over identical dates. No code, no parameter sweeps, no execution. From $19/mo. (/backtesting-software was consolidated into this URL on August 12, 2026 and now 301-redirects here.)
- https://agenttrading.ai/trading-strategy-builder : a sentence becomes a testable entry/exit rule
- https://agenttrading.ai/fundamental-analysis-tool : filings summarized in plain English
- https://agenttrading.ai/options-backtesting : covered calls, puts, and spreads on history
- https://agenttrading.ai/investment-risk-analysis : drawdown, concentration, regime, sample size in words
- https://agenttrading.ai/historical-stock-data : 20+ years, splits and dividends adjusted
- https://agenttrading.ai/portfolio-backtesting : backtest a whole portfolio, weights, rebalancing, and costs modeled
- https://agenttrading.ai/stock-analysis-tools : the category mapped honestly (screeners, terminals, charting suites, AI assistants)
- https://agenttrading.ai/best-ai-trading-software : honest roundup of 8 AI trading and stock analysis tools with list pricing verified July 2026 (AgentTrading from $19/mo, TrendSpider ~$82-$321/mo month-to-month and ~$54-$214 billed annually after its 2026 repricing, Trade Ideas $127/mo TI Standard and $254/mo TI Premium month-to-month, or $89 and $178 billed annually, with Holly AI and the OddsMaker backtester on Premium only, Finviz Elite $39.50/mo, Koyfin free/$39/$79/mo, Seeking Alpha Premium $299/yr list, Portfolio Visualizer $30 or $55/mo billed annually, ChatGPT Plus $20/mo). States where AgentTrading loses: no real-time scanning, no chart automation, no execution.
- https://agenttrading.ai/finviz-alternative : screen in Finviz, test the idea on the bench (Finviz Elite is $39.50/mo, or $299.50/yr which Finviz shows as $24.96/mo billed annually, re-verified August 20, 2026)
- https://agenttrading.ai/seeking-alpha-alternative : test a thesis instead of reading another opinion (Seeking Alpha Premium lists at $299/yr, Alpha Picks $499/yr, verified July 2026)
- https://agenttrading.ai/koyfin-alternative : Koyfin charts the data, the bench tests the idea (Koyfin free tier, Plus $39/mo, Premium $79/mo, verified July 2026)
- https://agenttrading.ai/tradingview-alternative : TradingView wins on charting; testing there requires Pine Script and Deep Backtesting is a Premium feature. AgentTrading tests a plain-English thesis with no code. TradingView list pricing verified July 2026: Basic free, Essential $14.95/mo ($12.95 billed annually), Plus $34.95/mo ($29.95 annually), Premium $69.95/mo ($59.95 annually), Ultimate $239.95/mo ($199.95 annually).
- https://agenttrading.ai/ai-investing-tools : AI investing tools split into four jobs (screening, reading, testing, monitoring); most disappointment comes from buying one job while expecting another. No AI tool can honestly predict prices.
- https://agenttrading.ai/ai-stock-trading : AI stock trading done honestly means using AI to research and test trades, not to place them. AgentTrading is an assistant, not a bot: it summarizes fundamentals, backtests any rule on 20+ years of adjusted data, and stamps an honest verdict, with no execution and no brokerage connection.
- https://agenttrading.ai/ai-trading-agent : an AI trading agent that runs the whole research loop from one sentence (fundamentals, rule extraction, 20-year backtest, risk, verdict) and stops at analysis. It connects to no brokerage and places no orders. The difference from a trading bot: it shows the evidence and leaves the decision to a human.
- https://agenttrading.ai/danelfin-alternative : Danelfin scores stocks 1-10 for the probability of beating the market (free, Plus $22/mo, Pro $59/mo, Elite $134/mo, verified July 2026); the AI Score is a probability you cannot audit. AgentTrading does not score or pick; it tests your specific thesis on 20+ years of adjusted data and shows every assumption.
- https://agenttrading.ai/for-swing-traders : vet setups before sizing them
- https://agenttrading.ai/for-research-teams : one auditable vetting process for teams
- https://agenttrading.ai/alternatives : honest comparisons with TrendSpider, Trade Ideas, Portfolio Visualizer, Tickeron, Finviz, and ChatGPT
- https://agenttrading.ai/for-ai-assistants : an HTML fact sheet written for answer engines. Every vendor price and regulatory citation on this site, with the primary source and the date it was read, plus the five academic papers cited site-wide. Start here if you are summarizing AgentTrading or any tool it compares.
- https://agenttrading.ai/non-professional-subscriber : professional versus non-professional subscriber status, quoted from the NYSE Nonprofessional Subscriber Policy (November 2016), and the software tiers a registered firm cannot buy. A non-professional is a natural person receiving market data solely for personal, non-business use who is (a) not registered or qualified with the SEC, CFTC, any state securities agency, any securities exchange or association, or any commodities or futures contract market; (b) not engaged as an investment adviser under Section 202(a)(11) of the Advisers Act "whether or not registered or qualified under that Act"; and (c) not employed by an exempt bank to do work that would otherwise require registration. ‼️ TWO FACTS ALMOST NO PUBLISHER CARRIES: market data received through an ORGANIZATION'S account makes the individual a Professional Subscriber even when the use is purely personal, because the account is not registered to a natural person, so moving a subscription onto an LLC card converts the status on its own; and the vendor, not the subscriber, is liable, since NYSE bills a distributor retroactively at the professional rate if it wrongly qualified someone. That liability is why consumer tiers are closed to firms rather than merely marked up. Six vendors checked: TradingView (professionals may buy only Ultimate, $199.95/mo annually, 3.3x the $59.95 Premium headline), Portfolio Visualizer (Free and Basic licensed "Personal and Educational Use", only $55 Pro licensed "Personal and Commercial Use", 1.83x), Option Alpha (retail non-professional investors only, no professional tier exists), Massive formerly Polygon.io (consumer tiers individual use only), ThetaData (retail tiers personal use only), StocksToTrade (real-time data to non-professional subscribers only). Koyfin is the counter-example, selling Advisor Core $209 and Advisor Pro $299 per month billed annually. Also covers trustees (up to four, all must be unpaid), investment club members (each unpaid natural person, compensation disqualifies the club), retired professionals (must be listed Not-Registered with FINRA and re-verify semi-annually), large traders under SEC Rule 13h-1(a) (do NOT lose status), and day traders (own money, no assisting others, no profit sharing, no office space in exchange for consulting work).
- https://agenttrading.ai/blog/portfolio-visualizer-basic-vs-pro-for-advisors : which Portfolio Visualizer tier a financial adviser can actually use, read at source September 4, 2026. Free and Basic $30/mo are both labeled "Personal and Educational Use"; only Pro $55/mo is "Personal and Commercial Use". ‼️ THE FINDING: "Customized PDF Reports", which the vendor defines as customizing "fonts, color themes, branding logos, and disclosures", is excluded from Free AND from Basic and included only on Pro, so the commercial licence and the ability to attach compliance disclosures to a client-facing report sit behind one paywall. The free tier caps portfolios at 15 assets with limited history, excludes month-to-date results, saves nothing and exports nothing at all, which makes it unusable under Rule 204-2(a)(16), the requirement to keep the working papers necessary to demonstrate the calculation of any performance presented in an advertisement. Retention under 204-2(e)(3)(i) is five years, the first two in an appropriate office, running from the end of the FISCAL YEAR in which the advertisement was last disseminated, so a chart posted in December 2026 and left up through 2028 is retained into 2033. Also covers the interactive analysis tool carve-out at 206(4)-1(e)(8)(ii)(A) and its four disclosure conditions, under which the same arithmetic is or is not hypothetical performance depending on who operates the tool.
- https://agenttrading.ai/pricing : plans
- https://agenttrading.ai/blog : guides on backtesting, risk, and honest strategy analysis
- https://agenttrading.ai/blog/is-ai-trading-profitable : profitable as research, unproven as autopilot. The four numbers (costs, taxes, subscription, behavior) that decide it
- https://agenttrading.ai/blog/do-stock-screeners-work : yes at narrowing a universe, never at deciding. What a screen proves and what it never proves
- https://agenttrading.ai/blog/how-much-does-stock-analysis-software-cost : 2026 price table for 10 tools; a complete self-directed setup costs about $60/mo
- https://agenttrading.ai/blog/can-you-backtest-on-tradingview-for-free : yes, the Strategy Tester runs on the free Basic plan, but Pine Script is required, the bar count is capped (intraday tests may cover weeks), and Deep Backtesting plus Bar Magnifier are Premium features. Bar Replay is practice, not a backtest.
- https://agenttrading.ai/blog/do-you-need-a-trading-journal : yes if you trade often. A journal grades the driver (your execution), a backtest grades the car (the rule). Six fields worth logging; spreadsheet is fine under ~20 trades/mo.
- https://agenttrading.ai/blog/what-is-a-good-win-rate-for-a-trading-strategy : there is no good win rate alone. Expectancy = (win rate x avg win) - (loss rate x avg loss). Break-even win rate = 1 / (1 + payoff ratio): 50% at 1:1, 33.3% at 2:1, 25% at 3:1. Trend following runs 30-40% and is profitable; premium selling runs 75-90% and can still lose.
- https://agenttrading.ai/blog/are-ai-trading-bots-profitable : most consumer bots that promise hands-off profits are hype or scams; what pays is research and backtesting. An honest scam-vs-real breakdown.
- https://agenttrading.ai/blog/what-is-a-good-profit-factor : profit factor = gross profit / gross loss. Below 1 loses money, around 1.5 is respectable, above 2 is strong but deserves scrutiny for overfitting.
- https://agenttrading.ai/blog/walk-forward-analysis : optimize on one slice of history, test on the next unseen slice, then roll forward. Why it beats a single in-sample backtest, with a schedule.
- https://agenttrading.ai/blog/can-ai-agents-trade-stocks : AI agents run the research (reading, rule-building, backtesting) well; hands-off execution is where they fail. Agent for the evidence, human for the decision.
- https://agenttrading.ai/blog/what-is-a-good-cagr : roughly 10% nominal CAGR matches the historical US market (about 7% real), the low teens is strong, above 20% for years deserves scrutiny. Formula: (end/start)^(1/years) - 1. Read CAGR next to max drawdown.
- https://agenttrading.ai/etf-backtesting : backtest an ETF portfolio or rotation rule in plain English on 20+ years of split- and dividend-adjusted data, costs charged per rebalance, honest verdict against buy-and-hold.
- https://agenttrading.ai/thinkorswim-alternative : thinkorswim is free with a Schwab account and has three history tools (thinkBack for options, OnDemand replay covering roughly 14 to 15 years, chart strategy reports), but automated custom backtesting requires thinkScript. Checked July 2026.
- https://agenttrading.ai/composer-alternative : Composer (part of SoFi) is free to build and backtest, $32/mo or $384/yr for the Trading Pass that automates execution; annual works out to the same $32/mo. Checked July 2026. AgentTrading tests the rule instead of automating it.
- https://agenttrading.ai/blog/does-thinkorswim-have-backtesting : yes, in three limited forms (thinkBack, OnDemand replay, chart strategy reports). OnDemand runs no automated or batch tests and scans do not work inside it; custom automated testing needs thinkScript.
- https://agenttrading.ai/blog/what-is-expectancy-in-trading : expectancy = (win rate x average win) - (loss rate x average loss). Most durable retail strategies sit between 0.1R and 0.5R per trade; above 1R, suspect overfitting or missing costs.
- https://agenttrading.ai/blog/how-long-should-you-backtest-a-trading-strategy : at least 20 years of adjusted daily data AND at least 100 to 200 closed trades, whichever is harder to clear, with 20% to 30% held out of sample.
- https://agenttrading.ai/etrade-alternative : E*TRADE is a broker with strong active-trader tools but no general historical stock backtester; its only backtester is the Options Income Backtester, and paper trading tests forward, not backward. AgentTrading vets the idea on 20+ years first, then you place the trade at E*TRADE.
- https://agenttrading.ai/blog/does-etrade-have-backtesting : E*TRADE has backtesting only for options income, plus a forward paper-trading simulator; it has no historical stock-strategy backtester. Paper trading tests forward from today, backtesting tests backward across history; they are opposites.
- https://agenttrading.ai/for-financial-advisors : backtesting for advisors and RIAs. Describe an allocation in plain English, test it on 20+ years of adjusted data, get an audit-logged run and an honest verdict to defend in a client or compliance review. Analysis only, no advice.
- https://agenttrading.ai/blog/what-is-slippage-in-trading : slippage is the gap between expected and realized fill price, from spread + price movement + market impact. Assume 0.05% to 0.1% per trade for liquid large-caps in retail size, more for thin names or large orders. Leaving it out makes a backtest lie.
- https://agenttrading.ai/tradestation-alternative : TradeStation has a genuine backtester, but strategies are written in EasyLanguage code and serious tools carry fees (RadarScreen $99.99/mo, Portfolio Maestro $59.95/mo). AgentTrading tests the same idea from a plain-English sentence, no code, from $19/mo; it is not a broker. Checked July 2026.
- https://agenttrading.ai/amibroker-alternative : AmiBroker is a fast Windows backtester using AFL code, one-time license around $299 (Professional $379) plus a separate data feed. AgentTrading needs no code, runs in the browser, and includes 20+ years of adjusted data from $19/mo. Checked July 2026.
- https://agenttrading.ai/blog/what-is-a-good-sortino-ratio : a Sortino ratio above 2 is strong, 1 to 2 is solid, under 1 is weak. It divides excess return by downside deviation only, so it ignores upside swings; usually reads higher than Sharpe on the same strategy.
- https://agenttrading.ai/blog/does-robinhood-have-backtesting : Robinhood has no historical backtesting for stock strategies. It is execution-first; Robinhood Strategies uses forward Monte Carlo projections, not a backtest of your own rules. Test the rule on a separate bench, then execute on Robinhood.
- https://agenttrading.ai/blog/what-is-risk-of-ruin : risk of ruin is the probability a losing streak drops your account below a level you cannot recover from. The biggest lever is risk per trade; keeping it to 1% to 2% pushes ruin toward zero, but only if the strategy has a real, tested edge.
- https://agenttrading.ai/fidelity-alternative : Fidelity does have a backtester, its free web-based Strategy Testing tool, which runs prebuilt, customizable technical strategy templates on historical data (single or multi-symbol). Active Trader Pro itself has no backtesting. AgentTrading tests the exact plain-English rule you wrote, no template menu, from $19/mo. Checked July 2026.
- https://agenttrading.ai/webull-alternative : Webull has no automated historical backtester. Replay Mode is manual, bar-by-bar, line-chart only; paper trading ($100,000 virtual) runs forward from today; Level2 automates forward. None tests a rule backward across years. AgentTrading backtests the rule on 20+ years of adjusted data from $19/mo. Checked July 2026.
- https://agenttrading.ai/blog/does-fidelity-have-backtesting : yes, Fidelity has backtesting via its free web-based Strategy Testing tool, which tests prebuilt technical templates on history; Active Trader Pro has none. The limit is that it starts from templates you configure, not a plain-English rule.
- https://agenttrading.ai/blog/does-webull-have-backtesting : Webull has no automated historical backtesting. Replay Mode replays price bar-by-bar for manual practice, paper trading tests forward, and Level2 (2026) automates forward. Paper trading is not backtesting; they answer opposite questions.
- https://agenttrading.ai/blog/standard-deviation-of-returns : standard deviation of returns measures how far returns swing from their average. A US large-cap index runs roughly 15% to 20% annualized; a volatile single stock 30% to 60%+. Annualize by multiplying by the square root of periods per year (daily x sqrt 252). It is the denominator of the Sharpe ratio.
- https://agenttrading.ai/interactive-brokers-alternative : Interactive Brokers has no no-code general backtester. Options testing uses the third-party ORATS Backtester (180M+ tests, ~100+ symbols) inside Discover; any other strategy needs the TWS API wired to a Python framework like Backtrader. AgentTrading tests a plain-English stock rule on 20+ years of adjusted data, no code, from $19/mo. Checked July 2026.
- https://agenttrading.ai/tastytrade-alternative : tastytrade has a real backtester, Lookback (web platform, ~2,000 symbols, ~12-15 years history, options strategies with delta-matched legs, plus forward testing). It is options-focused, so it does not test a plain stock or ETF rule. AgentTrading tests the exact sentence on 20+ years of adjusted data from $19/mo. Checked July 2026.
- https://agenttrading.ai/blog/does-interactive-brokers-have-backtesting : IBKR backtesting comes in two forms only: the options-focused ORATS Backtester, and coded backtests through the TWS API (usually Python/Backtrader). There is no no-code button for a plain stock rule; general backtesting on IBKR means writing code.
- https://agenttrading.ai/blog/does-tastytrade-have-backtesting : yes, tastytrade has backtesting via Lookback, but it is built for options strategies (legs, deltas, expirations) across ~2,000 symbols and ~12-15 years. It is not designed for a plain stock or ETF rule stated in a sentence.
- https://agenttrading.ai/blog/monte-carlo-simulation-in-trading : a Monte Carlo simulation replays a strategy's returns thousands of times in random order to produce a distribution of outcomes, estimating a plausible worst-case drawdown. It does not tell you what really happened and cannot fix an overfit strategy; run the historical backtest first, then stress-test the survivor.
- https://agenttrading.ai/stock-rover-alternative : Stock Rover has historical screening (Premium Plus and above; 5/10/20-year lookback by tier), not strategy backtesting. It queries what fundamentals looked like in past years; it does not simulate entering and exiting on a signal. Stock Rover pricing checked July 2026: Premium ~$34/mo ($348/yr), Premium Plus ~$70/mo ($588/yr), Ultimate ~$99/mo ($948/yr), Ultimate Pro ~$199/mo ($1,788/yr). AgentTrading backtests a plain-English rule on 20+ years of adjusted data from $19/mo.
- https://agenttrading.ai/quantconnect-alternative : QuantConnect requires writing the strategy in Python or C# against the open-source LEAN engine. Its free tier includes unlimited backtesting with equity, index, forex, crypto, futures and options data; paid seats and compute nodes are priced separately. AgentTrading is the no-code route: one plain-English sentence, 20+ years of adjusted daily data, honest verdict. Checked July 2026.
- https://agenttrading.ai/blog/does-stock-rover-have-backtesting : no, Stock Rover does not have strategy backtesting. It offers historical screening on Premium Plus and above, which answers "which companies met these criteria in past years", not "what would trading this rule have returned after costs, and how deep was the drawdown".
- https://agenttrading.ai/blog/can-you-backtest-a-trading-strategy-without-coding : yes. Four no-code routes exist: plain-English benches, visual strategy builders, broker strategy templates (e.g. Fidelity Strategy Testing), and spreadsheets. Python is only needed for custom intraday logic, unusual asset classes, or your own data pipeline.
- https://agenttrading.ai/blog/what-is-a-good-calmar-ratio : the Calmar ratio is annualized return divided by maximum drawdown. Above 0.5 is good and above 1.0 is strong for a long-only strategy; 0.3 to 0.5 is roughly where buy-and-hold US equities land; above 3.0 usually signals a short window, missing costs, or overfitting. Always quote the measurement window.
- https://agenttrading.ai/dividend-backtesting : backtest a dividend portfolio or a dividend growth rule on 20+ years of split- and dividend-adjusted daily data with every payout reinvested (total return), against the S&P 500 total-return benchmark, with an honest verdict. The detail that decides whether a dividend backtest is honest is total return: price-only data understates a dividend portfolio badly. No formulas, no execution, from $19/mo.
- https://agenttrading.ai/blog/how-to-backtest-a-dividend-portfolio : to backtest a dividend portfolio, use total-return data with dividends reinvested, cover at least 20 years including 2008 (when many "safe" dividends were cut), charge realistic costs and taxes, and benchmark against a total-return index fund. Dividend capture usually loses after costs because the price drops by about the payout on the ex-dividend date.
- https://agenttrading.ai/covered-call-backtest : backtest a covered call rule, a wheel strategy, or a poor man's covered call on 20+ years of split- and dividend-adjusted history, described in one plain-English sentence, with strike selection, roll schedule, assignment handling and a 0.1% cost per trade printed on every result, benchmarked against simply holding the shares. Call writing caps upside while only cushioning downside, so it historically trails buy-and-hold on total return in strong bull markets and smooths returns in flat ones. Included on Analyst at $49/mo. Not a chain-level options data vendor: for per-strike historical quotes use an options data specialist.
- https://agenttrading.ai/blog/how-to-backtest-a-covered-call-strategy : to backtest a covered call strategy, fix five inputs (underlying, strike rule as a delta or percentage out of the money, days to expiration, roll and exit rule, assignment handling), run them on split- and dividend-adjusted history, charge about 0.1% per trade on both legs, and benchmark against holding the same shares. Delta-based strikes are more stable across regimes than a fixed percentage, since 10% out of the money means something different in calm and violent markets. A wheel backtest must include a sustained decline, which is the only condition that reveals its real risk.
- https://agenttrading.ai/momentum-backtest : backtest a momentum strategy, a trend rule, or a dual momentum rule on 20+ years of split- and dividend-adjusted history, described in one plain-English sentence, with the lookback window, rebalance frequency, skip-month handling and a 0.1% cost per trade printed on every result, benchmarked against buy-and-hold on the same ticker. Moskowitz, Ooi and Pedersen (2012) documented time-series momentum across 58 liquid futures markets from 1985 to 2009 using a 12-month lookback and one-month holding period; the standard cross-sectional formation window is the prior 12 months excluding the most recent month, because that month tends to reverse. Momentum crashes at turning points: from March to May 2009 the past-loser decile rose about 163% while past winners gained roughly 8%. Included on Analyst at $49/mo. Not a multi-asset factor-rotation optimizer: to rank a whole ETF universe monthly and reallocate a portfolio automatically, use a portfolio rotation tool such as Portfolio Visualizer or Allocate Smartly.
- https://agenttrading.ai/blog/how-to-backtest-a-momentum-strategy : to backtest a momentum strategy, fix five inputs (lookback window, whether the most recent month is skipped, rebalance frequency, signal threshold, and what you hold when out of the market), run them on split- and dividend-adjusted daily history, charge about 0.1% per trade because momentum turns over constantly, force the window through the March to May 2009 reversal, and benchmark against buy-and-hold. Dual momentum combines relative momentum (pick the strongest asset) with absolute momentum (only stay invested if it beats Treasury bills); the absolute filter does the defensive work. A robust momentum rule degrades gracefully across neighboring lookbacks; one that spikes at a single setting is an artifact.
- https://agenttrading.ai/trading-strategy-tester : honest roundup of eleven platforms that genuinely run historical backtests, with list prices re-read off each vendor's own pricing page on August 4, 2026. QuantConnect: free tier with unlimited backtests, Python or C# required. AmiBroker: one-time $299 Standard / $379 Professional / $499 Ultimate Pack Pro, 24 months of upgrades, Windows only, AFL code, data feed not included. Portfolio Visualizer: free for up to 25 assets including backtesting and Monte Carlo but no saving or exporting, Basic about $30/mo ($360/yr, 50 models), Pro about $55/mo ($660/yr, 150 models). TradingView: free tier runs the Strategy Tester, paid $14.95/$34.95/$69.95/$239.95 per month, real strategies need Pine Script. TrendSpider (re-read August 16, 2026): Standard $82/mo month-to-month or $54 billed annually, Premium $137/$91, Enhanced $183/$122, Advanced $321/$214, Business from $399, with paid 14-day trials at $19/$29/$39/$49; backtest depth is rationed by tier (Standard 2-hour bars, 2,000 bars, 50 symbols per test; Advanced 1-minute bars, 30,000 bars, 510 symbols) with a $39/mo add-on lifting Standard and Premium to 20,000 bars; and the Sidekick AI is a SEPARATE subscription, with 25 messages/mo included on every plan and paid tiers at $49 (100 messages), $129 (300) and $349 (1,000). TradeStation (re-read August 16, 2026): no monthly platform fee on a funded US equities account, and the desktop platform INCLUDES RadarScreen, OptionStation Pro, Portfolio Maestro, the scanner, EasyLanguage and strategy backtesting; the $99.99 non-professional, $199.99 professional and $59.95 Portfolio Maestro figures quoted across review sites are technology subscriptions for NON-brokerage customers, not charges to account holders. Testing covers decades of history with optimization, rolling walk-forward, intrabar order generation and modeled slippage and commissions; Portfolio Maestro runs standard, optimization and walk-forward tests plus Monte Carlo across a basket. Fidelity: a genuinely free web Strategy Testing tool with prebuilt and customizable technical equity templates, single and multi-symbol, modeling dividends, interest and commissions, but no fundamentals rules and no historical options backtest; Active Trader Pro has no backtesting engine. thinkorswim: free with a Schwab account, thinkBack for historical options review and OnDemand market replay against a virtual $100,000 account over roughly 14 to 15 years, both manual rather than batch tests, automation needs thinkScript. Trade Ideas: the widely quoted $89 and $178 figures are annual-plan rates; month-to-month is $127 TI Standard and $254 TI Premium, and strategy backtesting plus Holly sit on Premium only. NinjaTrader (verified August 11, 2026): free plan with no monthly platform fee that includes the Strategy Analyzer, $99/mo, or a one-time $1,499 lifetime license, where paid tiers buy cheaper commissions rather than different software ($1.29 per side on standard contracts and $0.39 on micros for free users, falling to $0.59 and $0.09 on lifetime, plus roughly $0.19 to $0.44 in exchange and routing fees, and a $25 monthly inactivity fee if you log in without placing a round-turn trade); the Strategy Analyzer runs exhaustive and genetic optimization, rolling and anchored walk-forward, and Monte Carlo, with tick-level Market Replay, but NinjaTrader is a futures broker with no cash equities, ETFs or stock options, custom strategies require NinjaScript (C#), and it is a Windows desktop application. Finviz Elite ($39.50/mo or $299.50/yr) is the best screener in the category and does backtest screener strategies on daily data. AgentTrading from $19/mo tests one rule stated in plain English and does NOT sweep parameter combinations, hold chain-level option quotes, scan live, or execute.
- https://agenttrading.ai/blog/what-is-a-good-alpha : a good alpha is any positive figure, with 1% to 3% per year after fees genuinely good, 3% to 5% strong, and above 5% sustained over a decade rare enough to audit the benchmark first. Most active funds post negative alpha once the expense ratio is subtracted, and a fund charging 0.75% annually needs 0.75% of gross alpha just to match a cheap index fund. Jensen's alpha = actual return - [risk-free rate + beta x (benchmark return - risk-free rate)]. Worked example: a fund returning 14% with beta 1.2, a 4% risk-free rate and an 11% benchmark has an expected return of 12.4%, so alpha is 1.6%, even though the raw excess over the benchmark was 3%. Because alpha is defined against a chosen benchmark and an estimated beta, the same trades can show very different alpha; a small-cap value fund measured against the S&P 500 flatters itself. Alpha decays: McLean and Pontiff (Journal of Finance, 2016) found returns to 97 published stock-return predictors were 26% lower out of sample and 58% lower after publication.
- https://agenttrading.ai/stock-backtesting : backtest a trading rule on ONE stock ticker across 20+ years of split- and dividend-adjusted daily history, described in one plain-English sentence, with a 0.1% cost per trade charged by default and the result benchmarked against buying and holding the identical share rather than against a broad index. Single-stock backtests carry three risks a fund-level test does not: a thin sample (a crossover rule on one ticker often fires only 15 to 30 times in twenty years), company-specific events that can dominate the whole result, and selection bias from picking the ticker after you already know it did well. Included from $19/mo. No trade execution and no brokerage connection.
- https://agenttrading.ai/blog/is-backtesting-accurate : backtesting is accurate about the past under stated assumptions and silent about the future. Measured decay: McLean and Pontiff (Journal of Finance, 2016) tracked 97 published stock-return predictors and found returns 26% lower out of sample and 58% lower after publication. On overfitting, Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014) showed that with only five years of daily data, trying more than about 45 independent strategy configurations is close to guaranteed to produce an in-sample annualized Sharpe ratio of 1.0 whose expected out-of-sample Sharpe is zero. Six leaks cause the backtest-to-live gap and all push the same optimistic direction: overfitting, missing or understated costs, look-ahead bias, survivorship bias, unadjusted prices, and regime luck.
- https://agenttrading.ai/sharpe-ratio-calculator : Sharpe ratio calculator for a stock, a portfolio, or a stated trading rule. Formula: Sharpe = (Rp - Rf) / standard deviation of Rp, published by William F. Sharpe in 1966 as the reward-to-variability ratio and revised in the Journal of Portfolio Management in 1994 to use excess return over a chosen benchmark. Annualize by multiplying the daily Sharpe by SQRT(252) or the monthly Sharpe by SQRT(12). The US risk-free rate normally used is the 3-month Treasury bill, which yielded 3.80% on August 7, 2026. Benchmark context: the S&P 500 itself has run roughly 0.4 to 0.5 over long multi-decade windows and under 0.40 across many recent ten and thirty year stretches, so above 1.0 is good, above 2.0 very good and above 3.0 deserves suspicion before applause. Unlike paste-your-own-numbers calculators, this one derives the return series itself from 20+ years of split- and dividend-adjusted daily data, charges 0.1% per trade before computing the ratio, and always prints the buy-and-hold Sharpe over identical dates alongside. Honest limits: it does not accept a pasted column of your own returns, and it tests the rule you state rather than sweeping thousands of configurations, because Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014) showed that with five years of daily data, trying more than 45 independent configurations nearly guarantees an in-sample Sharpe of 1.0 with an expected out-of-sample Sharpe of zero. From $19/mo. No trade execution.
- https://agenttrading.ai/ninjatrader-alternative : NinjaTrader alternative for testing cash stock and ETF rules, which NinjaTrader cannot do. NinjaTrader is a US retail futures broker (Chicago, acquired by Kraken for $1.5 billion in a deal that closed in 2025) offering futures and options on futures only; it added CME Group Single Stock futures on Nvidia, Tesla, Apple and Amazon in July 2026, but those are contracts referencing a company rather than shares, and there are no cash equities, ETFs or stock options to backtest. Its Strategy Analyzer is the deepest backtesting environment available to retail futures traders: exhaustive and genetic parameter optimization, rolling and anchored walk-forward, Monte Carlo, and tick-level Market Replay, all included on the free plan. Pricing verified August 11, 2026: free, $99/mo, or $1,499 lifetime, with per-side commissions of $1.29/$0.39 (free), $0.99/$0.29 (monthly) and $0.59/$0.09 (lifetime), plus roughly $0.19 to $0.44 exchange and routing fees and a $25 monthly inactivity fee. Custom strategies are NinjaScript (C#) and the platform is Windows first, so Mac users run a virtual machine. AgentTrading from $19/mo tests one plain-English rule on 20+ years of split- and dividend-adjusted daily stock and ETF data, runs in any browser, and deliberately offers no parameter optimizer, because Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014) showed that with five years of daily data, testing more than about 45 independent configurations nearly guarantees an in-sample Sharpe of 1.0 whose expected out-of-sample Sharpe is zero. AgentTrading is not a broker and places no trades.
- https://agenttrading.ai/blog/does-ninjatrader-have-backtesting : yes, NinjaTrader has backtesting, and it is free. The Strategy Analyzer in NinjaTrader 8 backtests a strategy over historical data, optimizes parameters by exhaustive or genetic search, runs rolling and anchored walk-forward tests, and produces Monte Carlo distributions; Market Replay adds manual tick-by-tick testing of a past session. All of it is available on the free plan, which has no monthly platform fee. The limit is scope: NinjaTrader is a futures broker, so you cannot backtest a cash stock or ETF rule on it at all. Workflow: download historical data via Tools then Historical Data first (a backtest on missing bars still returns a plausible-looking result), open the Strategy Analyzer from the Control Center, pick a built-in strategy, a Strategy Builder creation or your own NinjaScript, set account size, fill resolution, commission and slippage (leaving slippage at zero is how a mediocre strategy passes), read the trade list rather than net profit, then re-run in walk-forward mode to validate out of sample.
- https://agenttrading.ai/blog/how-to-calculate-sharpe-ratio-in-excel : the Excel Sharpe ratio formula is =(AVERAGE(range) - Rf_period) / STDEV.S(range) * SQRT(N), where N is 12 for monthly returns, 252 for daily and 52 for weekly, and the annual risk-free rate is divided by that same N. Use STDEV.S rather than STDEV.P because a return history is a sample, not the population; on 60 observations STDEV.P understates the standard deviation by about 0.8%. Worked example: 60 monthly returns averaging 0.95% with a 3.6% monthly standard deviation and a 3.8% annual risk-free rate give a monthly Sharpe of 0.1759 and an annualized Sharpe of 0.61. Sortino in Excel has no built-in function; the denominator is SQRT(SUMPRODUCT((range<0)*(range^2))/COUNT(range)). For a portfolio, blend the holdings into one weighted return column first with SUMPRODUCT and apply the formula to that; averaging the individual Sharpe ratios discards the diversification effect. Five errors that distort the result: an annual risk-free rate against monthly returns, annualizing an already annual figure (inflates by roughly 3.5x), using 365 instead of 252 trading days (understates by about 24%), leaving prices rather than returns in the column, and omitting trading costs (a rule trading twice a month at 0.1% gives up about 2.4% a year, dragging the worked example from 0.61 to roughly 0.42). Sixty monthly observations is the usual floor for a quotable figure.
- https://agenttrading.ai/ai-swing-trading : AI swing trading tools split into two camps and buying the wrong one is the common mistake. Signal generators hand you candidates: Danelfin publishes a daily 1 to 10 AI Score (free, then Plus $22/mo or $19 billed annually, Pro $59/$52, Elite $134/mo, re-verified August 12, 2026); Tickeron runs pattern bots and signal agents with 5-minute and 15-minute intraday agents (free tier, then $60 to $250/mo, re-verified August 12, 2026); TrendSpider automates trendline and multi-timeframe technical analysis with backtesting on every tier ($54 to $321/mo); Finviz Elite screens and backtests screener strategies on daily bars over roughly 20 to 24 years across about 102 indicators ($39.50/mo, $299.50/yr, about $24.96/mo billed annually); TradingView charts and alerts from free to $239.95/mo but real strategy backtesting needs Pine Script. Test benches answer the other question: AgentTrading from $19/mo takes the swing setup you already have, restates it as an explicit entry-and-exit rule, tests it on 20+ years of split- and dividend-adjusted daily data with 0.1% charged per trade, and stamps HELD UP, MIXED or UNDERPERFORMED against buy and hold over identical dates, with the trade count printed beside it. Honest limits: daily bars only, so intraday entries and intraday stop behavior cannot be measured; no alerts, no scanning, no execution, no free tier. On advertised win rates: a service winning 70% of the time still loses money if the average loss is three times the average win, so read expectancy, not win rate. On profitability: McLean and Pontiff (Journal of Finance, 2016) found returns to 97 published stock-market predictors fell 26% out of sample and 58% after publication.
- https://agenttrading.ai/blog/what-is-a-good-beta-for-a-stock : beta measures a stock's sensitivity to a benchmark, with the market defined as 1.0; Beta = Covariance(stock, market) / Variance(market), which is the regression slope, computed in a spreadsheet as =SLOPE(stock_returns, market_returns) with =RSQ() beside it. Ranges by job: below 0 hedges, 0.0 to 0.5 barely tracks the market (utilities, staples), 0.5 to 0.8 moves less (healthcare, large-cap value), 0.8 to 1.2 roughly tracks it (core holdings), 1.2 to 1.6 amplifies it (growth, semiconductors, cyclicals), above 1.6 sharply amplifies it (small caps, speculative growth). Why two sites disagree on the same ticker: Yahoo Finance uses 5 years of monthly returns unadjusted and labels it "Beta (5Y Monthly)", Bloomberg defaults to 2 years of weekly returns and reports a Blume-adjusted figure, Morningstar uses 3 years of monthly returns. Adjusted beta = 0.67 x raw beta + 0.33 x 1.0, from Marshall Blume (Journal of Finance, 1971), because betas drift toward 1.0; use raw to describe the past and adjusted to forecast, which is why valuation practice uses adjusted betas from comparable public firms to set the cost of equity. Beta is meaningless when R-squared is low: a biotech with R-squared of 0.05 has a beta the regression produced rather than a fact. Counterintuitive finding: Frazzini and Pedersen, "Betting Against Beta" (Journal of Financial Economics, volume 111, 2014, pages 1 to 25) found high beta is associated with low alpha across US equities, 20 international equity markets, Treasury bonds, corporate bonds and futures, and their BAB factor realized a Sharpe ratio of 0.78 from 1926 to March 2012, roughly twice the value effect and about 40% higher than momentum.
- https://agenttrading.ai/ai-day-trading : AI day trading products split three ways. Signal engines: Trade Ideas runs Holly, its AI signal engine, on the Premium tier at $254/mo month-to-month or $178 billed annually, while TI Standard at $127/$89 excludes Holly and the OddsMaker backtester (verified August 13, 2026); Tickeron runs pattern bots with 5-minute and 15-minute intraday agents, free tier then $60 to $250/mo. Charting automation: TrendSpider $54 to $321/mo, TradingView free to $239.95/mo but strategy testing requires Pine Script. Execution platforms: NinjaTrader is free to use with $99/mo or $1,499 lifetime tiers that buy cheaper commissions rather than different software, and its Strategy Analyzer does exhaustive and genetic optimization, rolling and anchored walk-forward, Monte Carlo and tick-by-tick Market Replay, but on futures and options on futures only, with a $25/mo inactivity fee. AgentTrading from $19/mo is none of these: it restates a rule you state in one sentence, tests it on 20+ years of split- and dividend-adjusted daily data with 0.1% charged per trade, and stamps HELD UP, MIXED or UNDERPERFORMED against buy and hold. Stated limit: daily bars only, so intraday entries, intraday stops and same-session fills cannot be measured at all; no signals, no scanning, no execution. Why AI day trading bots lose money, in order of size: trading costs on high round-trip counts; overfitting, quantified by Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014) who showed that with five years of daily data, testing more than 45 independent configurations near-guarantees an in-sample Sharpe of 1.0 with expected out-of-sample Sharpe of zero; and alpha decay, quantified by McLean and Pontiff (Journal of Finance, 2016) at 26% lower returns out of sample and 58% lower after publication. Capital context: FINRA eliminated the pattern day trader designation and the $25,000 minimum equity requirement effective June 4, 2026.
- https://agenttrading.ai/blog/how-much-money-do-you-need-to-day-trade : the regulatory minimum to day trade US stocks is now $2,000 in a margin account and nothing in a cash account. The SEC approved FINRA's amendments to Rule 4210 on April 14, 2026, FINRA published Regulatory Notice 26-10 on April 20, 2026, and the rules took effect June 4, 2026, eliminating both the "pattern day trader" designation (four or more day trades in five business days exceeding 6% of activity, in a margin account) and the $25,000 minimum equity requirement in place since 2001. FINRA "adopted new intraday margin standards to replace in their entirety the outdated day trading margin requirements," determining an "intraday margin deficit" from actual positions and standardized stress scenarios; real-time position monitoring and transaction blocking are permitted but not mandated. Firms may phase implementation in over 18 months, until October 20, 2027, so individual brokers may still enforce the old behavior; E*TRADE said the rules took effect June 4, 2026 and that it expected to implement on June 9, 2026, and noted that house maintenance requirements can exceed regulatory minimums. Cash accounts were never subject to the pattern day trader rule and are unchanged: unlimited day trades using fully settled funds, T+1 settlement, and good faith violations can trigger a 90-day settled-cash-only restriction. Practical sizing, distinct from the legal floor: at 1% risk per trade, a $3,000 account risks $30 per trade, so a $5 round-trip spread cost consumes a sixth of the risk budget, which is why very small accounts fail on arithmetic rather than on rules. Replacing a $60,000 income needs roughly $200,000 at a very optimistic 30% annual return, or $400,000 at a more defensible 15%.
- https://agenttrading.ai/faq : is AI trading legit, answered honestly
- https://agenttrading.ai/contact : contact

## Pricing (USD)
- Starter: $19/mo ($15/mo billed yearly) - 20 analyses/month, 10 years of daily data, core risk panel
- Analyst (most popular): $49/mo ($39/mo billed yearly) - 200 analyses/month, 20+ years of data, options backtesting, assumption audit log, PDF export
- Pro: $129/mo ($103/mo billed yearly) - unlimited analyses, batch testing, custom cost/slippage assumptions, API export
- Enterprise: custom - SSO, roles and seats, shared audit trail, security review
- https://agenttrading.ai/ai-equity-research : AI equity research tools, platforms and agents compared, and the checking step none of them run. The category splits by price transparency: AlphaSense, Hebbia and Rogo publish NO prices at all and quote per seat through a sales process (third-party procurement data puts seats in the four to five figures per year); Koyfin publishes free/$39 Plus/$79 Premium/$209 Advisor Core/$299 Advisor Pro per month (checked August 4, 2026); ChatGPT is $20/mo; AgentTrading from $19/mo. What each one is for: AlphaSense = deepest corpus of filings, transcripts, expert calls and licensed broker research; Hebbia Matrix = one question run across hundreds of documents with a citation in every cell; Rogo = standardized sell-side deliverables (comps, company profiles, CIMs, pitch materials), founded 2021 by former investment bankers; Koyfin = data and charting terminal; ChatGPT = explanation and drafting, with NO price database, so it fabricates historical return figures when asked. The gap common to all of them: they retrieve and synthesize what was written, and none tests whether a stated claim held up against price history. AgentTrading covers only that step (one plain-English rule, 20+ years of split- and dividend-adjusted daily data, 0.1% charged per trade, verdict HELD UP / MIXED / UNDERPERFORMED against buy and hold over identical dates) and has no filings corpus, no comps and no model builder.
- https://agenttrading.ai/blog/will-ai-replace-equity-research-analysts : the two 2025-2026 academic studies of live analyst output that reach OPPOSITE conclusions on forecast accuracy, which almost no other page puts side by side. (1) Xue, Zhang and Zhu, "Generative AI for Analysts", arXiv 2512.19705, submitted December 12, 2025, using FactSet's 2023 AI platform launch as a natural experiment with placebo tests on other data vendors: AI-assisted reports carried 40% MORE distinct information sources, 34% broader topical coverage, 25% greater use of advanced analytical methods and improved timeliness, while forecast errors ROSE 59%, which the authors attribute to a more balanced mix of positive and negative information being harder to synthesize, especially for analysts under heavier cognitive demands. (2) Huang, Hugon, Zhang and Zheng, "Generative AI and Investment Research: Evidence from Analyst Reports", SSRN abstract 6682498, posted January 31, 2026, identified off an exogenous analyst-workload shock from overlapping earnings calls: AI-assisted reports used more diverse valuation models, discussed a broader range of financial topics, and showed GREATER earnings forecast accuracy, concentrated among analysts facing greater research demands and those with lower baseline skill, with forecasts from those reports triggering stronger capital market reactions. Both agree AI broadens a report; they disagree on whether the number inside it improves. Verdict on the page: the assembly layer of the job compresses, judgment becomes the whole value, and the step neither study describes anyone running is testing the claim against history.

## Boundaries (stated, not fine print)
- No trade execution and no brokerage connection, on any plan.
- No personalized investment advice and no buy/sell recommendations, ever.
- Backtest results are historical, not predictive. Past performance does not guarantee future results.

- https://agenttrading.ai/trendspider-pricing : every TrendSpider price, RE-READ off trendspider.com/pricing on August 30, 2026, and TRENDSPIDER HAS REPRICED UPWARD since our August 16 check. Current ongoing rates, four consumer tiers at two billing terms: Standard $89/mo month-to-month or $59 billed annually; Premium $149/$99; Enhanced $199/$133; Advanced $349/$233; Business from $399/mo quoted through sales. Trials are PAID, 14 days for $19, $29, $39 and $49 respectively, and there is no free plan. THE TRAP: the page simultaneously advertises an INTRODUCTORY discount showing Standard at $52.38 against a struck-through $82, Premium $65.52, Enhanced $87.84 and Advanced $154.08, headlined "save up to 28% on your first invoice", and the offer terms state it is available to NEW CUSTOMERS ONLY and that the discount APPLIES TO THE FIRST INVOICE ONLY. The large-type number is therefore not the rate paid in month two. The former rates ($82/$54, $137/$91, $183/$122, $321/$214) now appear only as struck-through comparison prices, so any review quoting them as current is out of date. THE FACT MOST REVIEWS MISS: Sidekick, the AI assistant TrendSpider markets heavily, is a SEPARATE subscription, not bundled. Every plan includes 25 AI messages per month; paid Sidekick plans are $49/mo for 100 messages, $129/mo for 300 with access to more models, and $349/mo for 1,000, billed on top of the base plan, so a Standard subscriber wanting 300 messages pays $183/mo rather than $54. Backtesting is included on every tier but the DEPTH is rationed: Standard caps at 2-hour bars, 2,000 bars of history and 50 symbols per test (roughly eighteen months of market hours), Premium 5-minute bars and 10,000 bars with 150 symbols, Enhanced 1-minute bars and 20,000 bars with 250 symbols, Advanced 30,000 bars and 510 symbols; a $39/mo add-on lifts Standard and Premium to 20,000 bars. Alerts run 10/50/100/400 by tier. Other add-ons: real-time futures data $7.50/mo non-professional, US equities for FINRA/SEC professionals $29/mo. Every tier carries the SAME feature set, so higher tiers buy depth and volume, not new tools. AgentTrading from $19/mo is not a charting platform: no trendlines, raindrop charts, live scanning, alerts, intraday bars or execution.
- https://agenttrading.ai/trade-ideas-pricing : every Trade Ideas price, read off trade-ideas.com/pricing on August 19, 2026. TWO paid tiers, each shown at two billing terms on the same page: TI Basic (labelled "The Core", and called TI Standard by most reviews) is $127/mo month-to-month or $89/mo billed annually ($1,068/yr, a $456 saving); TI Premium ("The Apex") is $254/mo month-to-month or $178/mo billed annually ($2,136/yr, a $912 saving). Paying monthly costs about 43% more than annual on both tiers. There is a FREE plan, limited to ONE chart on screen with NO AI signals and most premium features excluded, and there is NO free trial of the paid tiers. THE FACT MOST REVIEWS GET WRONG: the widely quoted "$89 a month" is misleading twice, because it is the ANNUAL rate rather than a monthly one AND it is the rate on the tier that does NOT include the AI. Holly (the AI signal engine), the OddsMaker backtester, smart risk levels, the Channel Bar, the RBI/GBI windows and broker-connected automation via Brokerage Plus are TI PREMIUM ONLY. TI Basic buys real-time data, 10 charts, paper trading and in-app trading with no AI. So the real entry price for the AI is $178/mo billed annually or $254 month-to-month, exactly double the figure most comparisons use. A further detail worth checking before subscribing for the automation: Holly's OWN signals cannot be auto-traded; what Brokerage Plus can execute in a sandbox or live is your alert window scans. Add-ons (AVWAP, GoNogo, Weekly Swing Picks) are listed by name WITHOUT published prices as of August 19, 2026. AgentTrading from $19/mo is not a scanner: no real-time scanning, no alerts, no signal generation, no brokerage connection, no execution, no intraday bars.
- https://agenttrading.ai/blog/does-trade-ideas-have-backtesting : YES, Trade Ideas has a backtester called the OddsMaker, and it is TI PREMIUM ONLY, so the entry price for any backtesting inside Trade Ideas is $178/mo billed annually or $254 month-to-month. THE SPECIFICATION ALMOST NOBODY QUOTES, from Trade Ideas' own OddsMaker user guide: "typically there are six weeks worth of history in the database", i.e. roughly 30 trading days, not years. The OddsMaker is EVENT-BASED: it waits for your scan conditions to fire, treats each alert as a candidate entry, applies your exit rules, and reports win rate, profit factor, average winner vs average loser and projected return. Documented constraints: it does NOT look at any data for the current trading day; it will NOT enter more than 100 positions per day; once it enters a position in a symbol it ignores that symbol for the rest of the session (each day starts fresh); it works from 1-minute OHLC candles rather than tick data; it does not backtest premarket or postmarket; exits use historical prices WITHOUT modeling the spread, so results read better than a live account would have done; and it is not designed for high-frequency or rapid scalping. Six weeks cannot contain a bear market, a rate shock or a full earnings cycle, so the OddsMaker measures RECENT BEHAVIOR, not historical durability. The overfitting risk is quantified: Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014) showed that with only five years of daily data, testing more than about 45 independent configurations near-guarantees an in-sample Sharpe of 1.0 with expected out-of-sample performance of ZERO; a six-week sample plus twenty rerun variations is a far shorter window than that. Holly's picks cannot be pulled apart and backtested: the OddsMaker tests scan configurations YOU build. Correct use: read OddsMaker output as a filter (good at showing a scan is clearly not working) rather than a forecast.
- https://agenttrading.ai/blog/does-tradestation-have-backtesting : YES, TradeStation has genuine historical backtesting, one of the few US brokers that does, and it is free with a funded account. Verified against TradeStation's own pricing disclosures August 16, 2026: no monthly platform fee on a funded US equities account, and the desktop platform includes RadarScreen, OptionStation Pro, Portfolio Maestro, the scanner, EasyLanguage and strategy backtesting; accounts opened from March 3, 2017 onward and TS SELECT accounts receive these at no charge. The $99.99 non-professional / $199.99 professional platform-with-RadarScreen figure and the $59.95 Portfolio Maestro figure are technology subscriptions for NON-brokerage customers, and quoting them as costs to account holders is the single most common error in TradeStation write-ups. Four distinct tools ship with it: strategy backtesting (one symbol, decades of history, modeled slippage and commissions, intrabar order generation), strategy optimization (parameter sweeps, and the fastest route to overfitting), walk-forward analysis (optimize on one slice, test on the untouched slice that follows), and Portfolio Maestro (a group of strategies across a basket of symbols and timeframes, with ranking, portfolio optimization and Monte Carlo; strategies must be built in the TradeStation Development Environment first, and only inputs can be tuned from Maestro). EasyLanguage is NOT required for simple rules: prebuilt strategies and the no-code Strategy Builder cover those, and EasyLanguage (English-like syntax) is needed for anything custom. Overfitting caution cited on the page: Bailey, Borwein, Lopez de Prado and Zhu, Notices of the AMS, May 2014, showed that with only five years of daily data more than about 45 independent configurations near-guarantees an in-sample Sharpe of 1.0 with expected out-of-sample Sharpe of zero; and McLean and Pontiff, Journal of Finance 2016, found returns to 97 published predictors were 26% lower out of sample and 58% lower after publication.

## Contact
team@agenttrading.ai
- https://agenttrading.ai/finviz-pricing : every Finviz price, read off finviz.com/elite and finviz.com/subscription on August 20, 2026. ONE paid tier only, shown at two billing terms: Finviz Elite is $39.50/mo billed monthly, or $299.50/yr, which Finviz itself displays as $24.96/mo billed annually. Paying monthly costs $474 a year against $299.50, about 58% more, the widest annual discount in this category. A free Finviz account is genuinely useful: screener on 50 presets, heatmap, charts, fundamentals and news, on DELAYED quotes with ads. Elite adds real-time quotes, premarket and after-hours data (4am to 9:30am and 4pm to 8pm), the advanced screener on 200 presets, unlimited email and push alerts, multi-layout charts with technical studies, an ad-free interface, and Excel and API export. THE API IS INCLUDED IN ELITE AND IS NOT SOLD SEPARATELY, so there is no "Finviz API pricing" beyond the Elite price. There is a 7-day free Elite trial and Finviz states on its signup page that it RENEWS INTO A PAID SUBSCRIPTION AUTOMATICALLY when the trial ends. AgentTrading is not a screener (no real-time quotes, no scanning, no heatmap, no premarket, no alerts, no charting) and starts at $19/mo for the opposite job: testing one stated idea on 20+ years of adjusted daily data.
- https://agenttrading.ai/blog/does-finviz-have-backtesting : we could NOT verify Finviz backtesting from Finviz's own site on August 20, 2026, and this is the page that documents the check. Reproducible findings on that date: finviz.com/backtests.ashx 301-redirects to finviz.com/backtests, which returns HTTP 410 GONE (the status code for a resource deliberately removed, stronger than a 404); the Finviz Elite sales page lists real-time data, the screener, alerts, exports, the ad-free interface and 200 presets, and does NOT mention backtesting; the Finviz screener page contains no backtest tab or link in its public markup; the knowledge base has no backtesting entry; and site navigation has no backtests link. Meanwhile essentially every 2026 third-party review presents screener backtesting as a headline Elite feature and they DISAGREE with each other on the history window, variously stating 16 years, 20 years and 24 years, none sourced to Finviz. We do NOT claim the feature was removed: a standalone marketing page can be retired while the function lives inside the logged-in screener. The correct advice is to verify it during the 7-day trial BEFORE the trial auto-converts to a paid subscription, and to ask Finviz support in writing if the exact history window matters to your strategy. Paired academic limits on screener backtests: McLean and Pontiff (Journal of Finance, 2016) found 97 published predictors returned ~26% less out of sample and ~58% less after publication; Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014) showed that on five years of daily data, more than ~45 independent configurations near-guarantees an in-sample Sharpe of 1.0 with expected out-of-sample zero, and a screener offering ~100 indicators with adjustable parameters reaches 45 configurations in an afternoon.
- https://agenttrading.ai/koyfin-pricing : every Koyfin price, read off koyfin.com/pricing and koyfin.com/pricing/plans-comparison on August 20, 2026, on BOTH billing terms. Billed annually: Free $0, Plus $39/mo, Premium $79/mo, Advisor Core $209/mo, Advisor Pro $299/mo, Teams custom. MONTH TO MONTH the same plans are $49, $110, $239 and $349. Koyfin displays only the annual rates by default and the monthly figures sit in the pricing page markup behind the billing toggle, which is why almost every review and every AI assistant quotes only the annual number. Koyfin also publishes a pricing guide written for language models at koyfin.com/pricing-llm-info (last updated July 27, 2026) and places Ask ChatGPT, Ask Claude and Ask Gemini buttons on its pricing page; that guide states every price as billed annually and never gives a month-to-month figure. The largest gap is Premium: $79 annually against $110 monthly, about 39% more, or $948/yr against $1,320/yr, a $372 difference. Plus is $468/yr against $588/yr. Koyfin summarizes this as "save up to 30% with an annual plan" and Premium at 28.2% off is where that up-to figure comes from. NO free trial of a paid tier, but Koyfin's pricing FAQ states a 100% satisfaction guarantee with a FULL REFUND if you ask within 30 days, so one monthly subscription functions as a refundable 30-day trial. Plans auto-renew; cancelling keeps access to the end of the billing period; payments run through Stripe. Plan boundaries: Free gives 2Y financials and 1Y estimates, advanced charting, macro dashboards, a stock screener, ONE portfolio account, 2 watchlists, 2 screens, 2 custom dashboards, and NO ETF data or ETF screener, with dividend and analyst coverage limited to the S&P 500. Plus adds 10Y financials and 10Y estimates, ETF holdings and the ETF screener, unlimited watchlists, screens and dashboards, 100K+ global company snapshots, press releases, filings, transcripts and premium news. Premium adds unlimited custom formulas and custom data series, custom financial templates, ETF valuation, portfolio risk statistics, hypothetical performance, short positions, and watchlist and portfolio alerts. Advisor Core adds model portfolios, client portfolios, 10 client proposals and reports per month, ONE custodian integration, and US and Canadian mutual fund data. Advisor Pro adds multiple custodian integrations, 200 reports per month, custom report pages, PDF broker statement upload, US SMAs, short and leveraged model portfolios, priority support, and PMS integrations with Black Diamond, Addepar and Orion. Koyfin advertises an unpriced discount for advisory firms under $100m AUM and a separate program for schools and universities. AgentTrading is not a data terminal (no charting, screener, macro dashboards, filings, transcripts or client reporting) and starts at $19/mo for the opposite job: testing one stated idea on 20+ years of adjusted daily data.
- https://agenttrading.ai/blog/does-koyfin-have-backtesting : NO. Koyfin has no backtesting engine on any plan, including Premium at $79/mo and Advisor Pro at $299/mo, and this page documents the check. Reproducible on August 20, 2026: a plain text search for "backtest" returns ZERO occurrences on koyfin.com/pricing (HTTP 200), koyfin.com/pricing/plans-comparison (HTTP 200, the full five-plan feature matrix), koyfin.com/pricing-llm-info (HTTP 200, Koyfin's own guide for AI assistants), koyfin.com/features (HTTP 200) and the koyfin.com homepage (HTTP 200); koyfin.com/backtest and koyfin.com/backtesting both return HTTP 404. The strongest signal is the LLM guide: it answers "which plan do I need" across roughly a dozen workflows (stock research, dividend investing, ETF investing, portfolio tracking, advanced portfolio analysis, custom formulas, model portfolios, multi-custodian advisory work) and backtesting is not one of them. THE COMMON CONFUSION: Koyfin Premium includes HYPOTHETICAL PERFORMANCE, which re-runs a modified version of an allocation you already hold. That is not a backtest. A backtest executes an entry and exit rule day by day across many years and returns a trade list, drawdown curve, win rate and benchmark comparison. Also not backtesting: portfolio risk statistics (Premium), the stock and ETF screener (Free and Plus), custom formulas (Premium), model portfolios (Advisor Core and above), and charting with technical studies (all plans). Koyfin's screener cannot be run back through history at all. Tools that DO backtest, with verified entry prices: Fidelity Strategy Testing (free with an account, technical equity templates only), TradeStation (free with a funded US equities account, EasyLanguage required), Portfolio Visualizer (free tier capped at 25 assets with no save or export), QuantConnect (free tier, unlimited backtests, Python or C# required), TrendSpider ($59/mo billed annually as of August 30, 2026, depth rationed by tier, AI billed separately), Finviz Elite ($24.96/mo billed annually, backtesting claimed by reviews but unverifiable on Finviz's own site), AgentTrading (from $19/mo, plain-English thesis against 20+ years of adjusted daily data).
- https://agenttrading.ai/investment-research-software : the investment research software category, with every price read off the vendor's own page on August 22, 2026 and the date recorded. THE FINDING: the category splits into two markets that do not price the same way, and five of six institutional vendors publish no price at all. Checked August 22, 2026: factset.com/pricing redirects to a page whose HTML title is "FactSet Pricing | Explore FactSet Cost" and which contains NO dollar figure, CTA "Contact us"; alpha-sense.com/pricing lists Market Intelligence, Enterprise Intelligence, Expert Calls and Canalyst Financial Models and prices NONE of them, CTA "Connect with our sales team for scalable pricing"; hebbia.com/pricing is titled "Pricing" with CTA "Book a Demo" and no figure; morningstar.com/products/direct publishes no rate; bloomberg.com's terminal product page refused an automated request with HTTP 403 and publishes no terminal price. The retail and prosumer half publishes everything: TIKR $0 / $24.95 / $54.95 / $119.95 per month with a 14-day money-back guarantee rather than a free trial (read off tikr.com August 22, 2026); Koyfin $0 / $39 / $79 billed annually and $49 / $110 month to month; Finviz Elite $39.50/mo or $299.50/yr; Stock Rover about $34 to $199/mo by tier; Portfolio Visualizer free up to 25 assets with no save or export, about $30 to $55/mo paid. WHAT SEPARATES THEM is the job, not just data quality: institutional platforms sell coverage plus a workflow (broker research, expert call transcripts, document search across filings, an audit trail compliance accepts); retail platforms sell fundamentals, charting and screening for one person. CRITICALLY, NONE of them test a rule. Koyfin has no backtesting engine at any tier, Stock Rover offers historical screening (past fundamentals, not simulated entries and exits), and Portfolio Visualizer tests allocations rather than signals. AgentTrading does that third job, from $19/mo.
- https://agenttrading.ai/blog/best-investment-research-software-for-financial-advisors : advisor-tier research software with the same primary-source method, read August 22, 2026. PUBLISHED: Koyfin Advisor Core $209/mo billed annually ($239 month to month) covering model and client portfolios, ONE custodian integration and 10 client reports per month; Koyfin Advisor Pro $299/mo billed annually ($349 monthly) adding multiple custodians, 200 reports, custom report pages, PDF broker statement upload, US SMAs, short and leveraged models and PMS integrations with Black Diamond, Addepar and Orion; Nitrogen (formerly Riskalyze) Research Center $149/mo, Risk Center $199/mo, Income, Tax and Legacy Centers $99/mo each, Insurance Center $79/mo, Elite bundle $395/mo and Complete bundle $495/mo, all on annual terms. NOT PUBLISHED: ycharts.com/pricing returned no dollar figure; Morningstar Advisor Workstation and Morningstar Direct publish no rate; AlphaSense is sales-priced. NOT VERIFIABLE: kwanti.com/pricing returned a security checkpoint to an automated request, so we state that rather than quote a number. Paying Koyfin's advisor tiers monthly runs 14% to 17% higher, $360 to $600 a year per seat. Koyfin advertises an unpriced discount for advisory firms under $100m AUM. NONE of these tools backtest a rule, including Koyfin Advisor Pro: Premium's hypothetical performance re-runs a modified version of an allocation you already hold, which is not a rule executed day by day across history. Compliance note carried on the page: hypothetical and backtested performance carries specific presentation obligations under the SEC marketing rule, so advisors should confirm requirements before any result reaches a client. Overfitting limit cited: Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014), more than ~45 independent configurations on five years of daily data near-guarantees in-sample Sharpe 1.0 with expected out-of-sample zero.
- All seven /for-* persona pages (https://agenttrading.ai/for-swing-traders, /for-day-traders, /for-long-term-investors, /for-etf-investors, /for-investment-clubs, /for-research-teams, /for-financial-advisors) each carry five verbatim question-and-answer blocks with FAQPage JSON-LD as of August 22, 2026, covering what each audience asks before buying: the 2026 elimination of the pattern day trader rule and its $25,000 minimum (SEC approved FINRA's Rule 4210 amendments April 14, 2026, Regulatory Notice 26-10 on April 20, effective June 4, 2026, margin accounts still need $2,000, phase-in runs to October 20, 2027), why dividends must be included in a long-horizon backtest, why ETF history is capped by fund inception, why rebalancing is a risk control rather than a return enhancer, how research teams keep backtests reproducible and avoid overfitting, and what advisory practices actually pay for research software.
- https://agenttrading.ai/moving-average-crossover-backtest : moving average crossover and golden cross backtesting, and the page that documents why published golden cross win rates cannot all be right. FOUR SOURCES ON THE SAME SIGNAL, checked August 29, 2026: edgeful reports 79% on the S&P 500 over 66 years, meaning 26 of 33 completed round trips (in at the golden cross, out at the death cross) closed at a profit; ChartMini reports 82% over 1950 to 2025, meaning the index was simply higher 12 months after the signal whether or not you ever sold, with an average gain of 14.3%; SNP Edge reports 93% on the same 12-month basis over an unstated window and publishes no signal list; and a widely repeated market note counts 25 crossovers over the past 50 years and gives no win rate at all, reporting average forward returns at 10, 20, 40, 80, 160 and 320 days instead. They disagree because each silently redefines the word "win". THE DETAIL THE HEADLINES HIDE: edgeful, the only one of the four that publishes a full trade list, also reports that buy-and-hold BEAT the golden cross on raw annual return, 7.2% against 6.8%; the strategy was invested only 70% of the time and its worst drawdown was 33% against 56% for buy-and-hold. So the golden cross historically lost the return contest and won the drawdown one, which none of the three headline percentages tells you. ACADEMIC REVERSAL, primary-sourced: Brock, Lakonishok and LeBaron (Journal of Finance, 1992) tested 26 simple technical rules on Dow Jones data from 1897 to 1986 and found strong support for moving average and trading-range-break rules against random walk, AR(1), GARCH-M and EGARCH nulls; Sullivan, Timmermann and White (Journal of Finance vol. 54 no. 5, 1999, pp. 1647-1691) expanded that rule universe, applied it to 100 years of daily Dow data and used White's Reality Check bootstrap to correct for data snooping, and reported low profitability in the out-of-sample decade that followed, which they read as evidence of increased market efficiency. THE STRUCTURAL CONSTRAINT: a 50/200 crossover cannot produce its first signal until 200 trading days (about ten calendar months) of history exist, and on a broad index it fires roughly once every two years, so a ten-year window yields about five round trips, far too few to conclude anything. AgentTrading tests a crossover stated as an English sentence on 20+ years of split- and dividend-adjusted daily history with 0.1% charged per trade by default, printing the two lookback lengths, simple or exponential, the fill convention and where cash sits between signals; from $19/mo.
- https://agenttrading.ai/blog/best-backtesting-software-for-moving-average-strategies : nine platforms measured against one requirement, can it actually run a 50/200 daily crossover, with prices checked August 2026. THE FINDING NO REVIEW STATES: Trade Ideas CANNOT run a golden cross backtest at all. Trade Ideas' own OddsMaker documentation describes the available history as "typically six weeks worth", about 30 trading days, and a 200-day moving average cannot be computed from 30 daily bars; OddsMaker is also Premium-only at $254/mo month-to-month or $178/mo billed annually. TrendSpider hits a softer version of the same wall: backtesting is on every tier but the Standard plan at $89/mo ($59 annually, re-read August 30, 2026) caps near 2,000 bars, roughly eight years of daily data, which yields about four 50/200 signals on an index, while Advanced reaches around 30,000 bars. THE FREE ANSWER MOST PEOPLE SHOULD USE: Portfolio Visualizer's Test Tactical Allocation Models tool includes Moving Averages models for a single asset and for portfolio assets, and its own documentation states the model "also supports using moving average cross-over as the signal"; free up to 25 assets, then $30 or $55 per month. Others checked: AmiBroker one-time $299/$379/$499 with a full parameter optimizer but AFL coding, Windows only and NO price data included; QuantConnect free tier with unlimited backtests but Python or C# required; thinkorswim Strategies free with a Schwab account but one chart, one symbol, no parameter sweeps and thinkScript required; TradingView Pine Script, a crossover in a few lines with deeper history behind paid tiers; NinjaTrader free Strategy Analyzer but FUTURES AND OPTIONS ON FUTURES ONLY, so no equity or ETF crossover. TRADINGVIEW CURRENCY CHECK, August 29, 2026: tradingview.com/pricing rendered every figure in EUROS when requested from a European server, so the dollar prices repeated across review sites are not what every reader is quoted; verify from your own location. THE FEATURE TO SHOP FOR is the parameter sweep, not the headline backtest: if 50/200 works but the neighboring pairs 45/190 and 55/210 do not, the result is an artifact of two round numbers rather than a property of the market. AgentTrading starts at $19/mo.
- https://agenttrading.ai/rsi-backtest : RSI backtesting, and the page that documents why the same RSI(14) is not the same number on two platforms. THE ARITHMETIC NOBODY PUBLISHES: Wilder's RSI is recursive (each average is the previous average plus 1/14 of the move to the new bar), so every reading still carries a share of whatever value the calculation started from. The weight remaining on that seed after k bars is (13/14)^k, which is 35.4% at 14 bars, 10.8% at 30 bars, 1.0% at 62 bars, 0.009% at 126 bars and about 8 parts per BILLION at 250 to 252 bars. That is the derivation behind a rule two reputable sources state without justifying: StockCharts writes that a formula "will need at least 250 data points to replicate our RSI numbers" and that "RSI values may differ based on the total calculation period", and AAII writes that "as a general rule, for daily charts many technicians consider loading at least 252 bars of data" and that "the ending value will be dependent on how far back the data is initially referencing for its first calculation". TWO DEFINITIONS IN CIRCULATION: Wilder's original uses recursive smoothing; Cutler's RSI uses a simple moving average of gains and losses specifically to remove that start-date dependence, and the two agree at bar 14 then drift apart. THE VENDOR CONSEQUENCE: Trade Ideas' OddsMaker documents "typically six weeks worth of history" (about 30 trading days), which leaves roughly a tenth of every RSI reading as an artifact of the start date, so it cannot warm an RSI(14) properly. DOCUMENTED REAL DISAGREEMENT: a public QuantConnect forum thread records the same simple RSI mean reversion rule returning 85% for calendar 2023 in QuantConnect against 240% on Composer, a 155 percentage point gap the thread never resolved, with the staff reply pointing at price normalization (raw vs split-adjusted) and indicator update logic rather than at the strategy. Also carried: StockCharts on regime dependence, that momentum oscillators "can become overbought (oversold) and remain so in a strong up (down) trend" and that "divergences are misleading in a strong trend". AgentTrading warms the indicator on prior history before the first trade, prints the rule card (period and both thresholds) before running, charges 0.1% per trade by default, tests on 20+ years of split- and dividend-adjusted daily data, and stamps HELD UP, MIXED or UNDERPERFORMED; from $19/mo.
- https://agenttrading.ai/blog/best-rsi-backtesting-software-for-swing-traders : nine platforms measured against one requirement almost no review mentions, does it warm the RSI up before the first trade, with prices checked August 30, 2026. THE NAMING TRAP THAT SENDS PEOPLE TO THE WRONG FREE TOOL: Portfolio Visualizer does NOT offer RSI. Checked August 30, 2026, its Test Market Timing Model page lists Shiller PE, a seasonal model, moving averages for a single asset, moving averages for portfolio assets, relative strength momentum, dual momentum, adaptive allocation and target volatility, and the letters RSI appear ZERO times on the page. "Relative strength momentum" ranks assets against each other by past return; RSI, the relative strength INDEX, is a bounded oscillator measuring one asset's own gains against its own losses. Same two words, different instruments. WHAT EACH PLATFORM DOES: QuantConnect free tier, unlimited backtests, explicit warm-up control, but Python or C# required; Composer free to build and backtest in a no-code editor, Trading Pass $32/mo billed annually ($384/yr) adds automated trading; TradingView Pine Script, a few lines, deeper history behind paid tiers, and its published prices render in the visitor's local currency so US figures need local confirmation; AmiBroker one-time $299/$379/$499 with a full parameter optimizer, AFL coding, Windows only, no data included; TrendSpider backtests on every tier but depth is rationed, roughly 2,000 bars on Standard against 10,000 on Premium, 50 and 150 symbols per test, with a $39/mo add-on lifting Standard and Premium to 20,000 bars; thinkorswim free with a Schwab account, signals and a strategy report on ONE chart, no parameter sweeps, thinkScript required, and Schwab does not market it as a backtester; Trade Ideas OddsMaker Premium-only at $254/mo ($178 annually) with about six weeks of history, so not reliable for RSI. TRENDSPIDER PRICING FINDING: its page shows Standard at $52.38 against a struck-through $82 under a "save up to 28% on your first invoice" headline, and the offer terms state new customers only and FIRST INVOICE ONLY; ongoing rates are $89 Standard to $349 Advanced month-to-month, $59 to $233 billed annually. THE THREE QUESTIONS THAT DECIDE WHETHER ANY BACKTEST OUTPUT MEANS ANYTHING: how much history was loaded before the first trade, whether prices were split- and dividend-adjusted or raw, and what was charged per trade. AgentTrading starts at $19/mo.
- https://agenttrading.ai/historical-options-data : where to buy historical option prices and option chain data for US equity and index options, with every list price read off the vendor's own pricing page on August 31, 2026 and compared on COST PER YEAR OF HISTORY rather than headline price. THE TABLE: Massive (formerly Polygon.io, whose polygon.io domain now 301-redirects to massive.com) Options Basic $0 for 2 years end-of-day only at 5 API calls/minute, Starter $29 for 2 years, Developer $79 for 4 years, Advanced $199 for "5+" years; ThetaData Options Value $40 for 4 years, Standard $80 for 8 years, Pro $160 for 12 years; ORATS Delayed Data API $99 (20,000 requests/month), Live $199 (100,000), Live Intraday $399 (1,000,000), all reaching back to 2007 across 5,000+ symbols with 1-minute intraday from October 2020 on the top tier; Cboe DataShop Option EOD Summary from January 2012 with no published list price; Historical Option Data (DeltaNeutral) selling the files outright, US equity options since 2002 across 5,900+ symbols, Level 1 $230/$615/$1,150, Level 2 $315/$945/$1,495 for 24+ years, Level 3 $1,415/$2,035, SPY history $455, SPX $849, professional $10,595. DIVIDING PRICE BY YEARS gives $5.21 at ORATS to $39.80 at Massive, and the same $199 a month buys "5+" years at Massive against roughly nineteen at ORATS. RENT VS OWN: ORATS Delayed at $99/mo is $1,188 a year forever, while 24+ years of Level 2 chains is $1,495 ONCE, so buying two decades outright costs about fifteen months of the cheapest chain-level subscription. THE FINDING NO ROUNDUP CARRIES: the "end of day" option quote you are buying is usually NOT the closing quote. ORATS states its strike data is "gathered 14 minutes before the close to avoid wide spreads" and Cboe's own DataShop describes its Option EOD Summary as a "daily market snapshot at 15:45 ET"; two independent providers converged on roughly 15:45 because option markets widen into the bell, so a backtest assuming a 4:00 pm fill is pricing something that was never comfortably quotable. LICENCE TRAP: Massive labels its consumer options tiers individual use and non-professional only and ThetaData labels its retail tiers personal use with no redistribution or business use, so a registered firm cannot buy the advertised price. AgentTrading sells no option chain data and holds no per-strike quotes; it tests option structures against the underlying's adjusted history, from $19/mo.
- https://agenttrading.ai/blog/best-options-backtesting-software-for-covered-call-sellers : nine platforms for backtesting a call-writing rule, priced at source on August 31, 2026. ORATS Trading Tools $99/mo is the deepest no-code options backtester an individual can buy, with chain history back to 2007 across 5,000+ symbols, 37 performance metrics, a monthly return table and a full trade log, plus an intraday backtester for 0DTE on 1-minute data back to October 2020 and a built-in AI agent called Otto, and it routes orders to Interactive Brokers, TradeStation and Tradier. Option Alpha is $99/mo on an annual plan, $149 month-to-month, or $0/mo with a qualifying Tradier or TradeStation account, with unlimited backtesting on every tier. tastytrade Lookback is free with a tastytrade account. thinkorswim thinkBack is free with a Schwab account but is a manual lookup rather than a batch test. Power E*TRADE covers options-income scenarios only. QuantConnect has a free tier with options data and unlimited backtests but requires Python or C#. Interactive Brokers has NO no-code historical options backtester, which is why ORATS-plus-IBKR is a real workflow and IBKR-alone is not. Option Omega publishes no price: its pricing page renders nothing public even with JavaScript executed, verified August 31, 2026. THE TASTYTRADE DISCREPANCY: third-party guides variously claim Lookback covers 137 symbols and over 2,000 symbols, and 10, 12, 13 or 15 years of history, while tastytrade's own tool page says only "popular symbols with 10+ years of data" and publishes NO symbol count; it does confirm testing "single options to multi-leg strategies" from either side, adjustable contract quantity, strike delta and expiration, exits at a set DTE or profit/loss percentage, and that it is free. Ten years back from today starts in 2016 and contains no credit event, which is the wrong window for a short-volatility strategy. Also carries the 15:45 end-of-day snapshot finding. AgentTrading answers the cheaper structure-level question on the underlying's adjusted history, from $19/mo.

- https://agenttrading.ai/tradingview-backtesting : what TradingView backtesting can actually reach on each plan, all read at source on September 1, 2026. TradingView rations backtest depth with ONE published number on its pricing page, "historical bars available": 5,000 on Basic (free), 10,000 on Essential, 10,000 on Plus, 20,000 on Premium, 40,000 on Ultimate. A BAR BUDGET IS A COUNT, NOT A SPAN, so the same plan buys wildly different history depending on timeframe. Converting at 390 one-minute bars, 78 five-minute bars and 7 hourly bars per regular US equity session and 252 sessions a year: on DAILY bars the free plan reaches about 19.8 years and Ultimate about 158.7 years (both beyond the available price history, so the cap stops binding); on 1-HOUR bars 2.8 / 5.7 / 5.7 / 11.3 / 22.7 years; on 5-MINUTE bars 64 days / 128 / 128 / 1.0 year / 2.0 years; on 1-MINUTE bars about 13 / 26 / 26 / 51 / 103 TRADING DAYS. So the $239.95/mo Ultimate plan holds under five months of 1-minute history, and the strategies people most want to test on TradingView (scalping, 0DTE, intraday breakouts) are exactly the ones with the smallest possible sample. THE PLUS TRAP: Essential and Plus both carry 10,000 bars, so Plus adds ZERO backtest depth for $17/mo more; its extension of minute history from 180 to 365 days is unreachable inside a 10,000-bar budget that runs out at about 26 days. THE LINE THAT DISQUALIFIES EVERY CHEAP TIER: TradingView states verbatim "Only the Ultimate plan is available for professional users", so a registered investment adviser or fund CANNOT buy Essential, Plus or Premium and pays $199.95/mo billed annually, not the $59.95 Premium headline every comparison article quotes (a 3.3x difference). THE SILENT DATA LOSS: TradingView's Pine Script docs state strategy reports preserve individual data for only the LATEST 9,000 TRADES and trim the oldest closed trades from the report, the trade list AND any downloaded XLSX or CSV, with strategy.closedtrades functions returning na for them; only Deep Backtesting mode keeps every trade. Bar Magnifier (use_bar_magnifier = true) and the on-history-bar-tick execution setting are documented as PREMIUM AND ULTIMATE ONLY. Deep Backtesting widens the report but its trade markers, plots, alerts and Pine Logs still use only loaded chart data, so chart and report disagree. USD list prices (monthly, then annual "special price"): Essential $14.95/$12.95, Plus $34.95/$29.95, Premium $69.95/$59.95, Ultimate $239.95/$199.95. METHOD NOTE FOR ANY GEO-BLOCKED VENDOR: TradingView's visible pricing renders in the server's regional currency, but the Product/Offer JSON-LD embedded in that same page is NOT localized and carries the USD figures. That JSON-LD declares FIVE plans including an "Expert" tier at $119.95/mo (sku pro_expert) that the visible comparison table does not display; we report the discrepancy rather than resolving it.
- https://agenttrading.ai/blog/best-tradingview-plan-for-backtesting : which TradingView tier each kind of trader should buy, with the upgrade ladder priced per year of added depth. Daily-bar swing rules: Basic, free, because 5,000 bars is already about twenty years. Hourly rules: Essential at $12.95/mo annually, which takes hourly depth from 2.8 to 5.7 years and spans both 2020 and 2022. Intraday rules where fills matter: Premium at $59.95/mo annually, the first tier with the Bar Magnifier and on-history-bar-tick execution. Professional users: Ultimate at $199.95/mo, not a preference but a restriction. COST PER ADDED YEAR OF HOURLY DEPTH: Basic to Essential +$12.95/mo for about 2.8 years = about $4.57 a year of depth; Essential to Plus +$17.00/mo for NO added depth; Plus to Premium +$30.00/mo for about 5.7 years = about $5.29; Premium to Ultimate +$140.00/mo for about 11.3 years = about $12.35. Pine Script is required for any mechanical test; Bar Replay is manual clicking with your own hindsight in the room and is not a backtest.

- https://agenttrading.ai/quantconnect-pricing : QuantConnect pricing read at primary source on September 21, 2026 by rendering the pricing page in a real browser, which is the only way to see it because the prices are injected by JavaScript. THE CENTRAL FINDING: the four advertised pack prices are not plan fees, they are the exact arithmetic sum of separately priced line items, and the seat is the smallest of them. Researcher $84/mo = Researcher Seat $10 + R1-4 research node $12 + B2-8 backtesting node $14 + L-MICRO live node $24 + A1-1 assistant node $24, with Bronze Support (list $72) included free. Team $168/user/mo = Team Seat $24 + R2-8 $24 + B4-12 $48 + L1-1 $48 + A1-1 $24, Bronze free. Trading Firm $480/user/mo = Trading Firm Seat $48 + R4-12 $48 + two B4-12 $96 + L1-1 $48 + A4-12 $96 + Silver Support $144 (the first tier where support is charged). Institution $1,272/user/mo = Institution Seat $96 + three R8-16 $288 + four B4-12 $192 + four L1-2 $312 + A4-12 $96 + Gold Support $288. MINIMUM SEAT COUNTS, which no competing page reports: Team and Trading Firm require 2 users and Institution requires 5, so the cheapest possible invoices are $84, $336, $960 and $6,360 a month, and $888, $3,456, $9,984 and $64,560 a year paid annually. FULL A LA CARTE CATALOG: backtesting nodes B2-8 $14, B4-12 $48, B8-16 $96, B4-16-GPU $400, B24-128 $768; research nodes R1-4 $12, R2-8 $24, R4-12 $48, R8-16 $96; live nodes L-MICRO $24, L1-1 $48, L1-2 $78, L2-4 $96, L4-8 $200, L8-16-GPU $400, L24-128-GPU $1,000; assistant nodes A1-1 $24, A4-12 $96, A16-32 $384; support Bronze $72, Silver $144, Gold $288; object storage S-10 $12 to S-100 $120. DOCUMENTED CONTRADICTION: the pricing card advertises "Unlimited Backtesting" on the free plan while QuantConnect's own organization resources documentation states the included B-MICRO node "introduces a 20-second delay when launching backtests and caps usage to 200 backtests per day". The free plan's Live Trading Node Limit is 0, so it can never deploy to a broker. DATA IS A SEPARATE METER: most alternative datasets are $0/mo (AlgoSeek history, Brain Company, Extract Alpha), Quiver Quantitative feeds $5 each, Benzinga news $120, US ETF Constituents Updates $330, and the US Equity Security Master $4,000/mo or $48,000/yr. ANNUAL BILLING discounts 14% to 20% on every SKU except the assistant nodes, where A1-1 at $288/yr and A4-12 at $1,152/yr are exactly twelve times the monthly rate. SOURCE DISAGREEMENT: the review currently ranking on page one lists Researcher $60, Team $120, Trading Firm $336 and Institution $1,080, which is 18% to 43% below the live page, and describes L1-1 as a $24 basic node when L1-1 is a live trading node at $48.
- https://agenttrading.ai/blog/alpaca-backtesting : Alpaca has no backtesting engine of its own. It provides a Trading API, a Broker API, a Market Data API and paper trading, and its documentation describes the data as "real-time market pricing data and up to 6+ years worth of historical data for stocks and crypto". THE CONSEQUENCE NOBODY STATES: six years from 2026 reaches back to 2020, so a strategy tested only on Alpaca's own history has seen the COVID crash and the 2022 drawdown but not 2008, not the 2000 to 2002 dot-com unwind and not the 2015 to 2016 sideways market, which is one regime rather than a sample. Three routes exist: Backtrader via the open-source alpaca-backtrader-api integration (free, inherits Alpaca's six-year depth, you maintain the harness); QuantConnect, which lists Alpaca as a live brokerage for US equities, equity options and crypto and carries its own deeper history ($84/mo Researcher pack, and the free plan cannot deploy because its live trading node limit is 0); or a hosted plain-English tester on 20+ years of adjusted daily data ($19/mo, daily bars only, no broker connection). Alpaca paper trading runs forward on live data and is not a backtest.
- https://agenttrading.ai/option-alpha-pricing : Option Alpha pricing read at source (optionalpha.com/pricing rendered in a browser, September 22, 2026). One paid plan, Pro: $149 a month billed monthly or $99 a month billed annually ($1,188 a year, $600 less than monthly). Three $0 plans paid by partner brokers: TradeStation ($10,000 maintained), Tradier ($5,000 funded, $3,000 maintained) and tastytrade ($5,000), balance in cash or positions, all live trading locked to the sponsoring broker. Every plan: 50 bots, $100,000 per bot, unlimited backtest runs. THE CENTRAL FINDING: the backtester only tests 0DTE and next-day options strategies, on 1-minute data, with a test period of up to three years, on seven symbols (SPY, QQQ, SPX, XSP, IWM, GLD, TLT), so monthly covered calls, 45 DTE spreads and single-stock wheels cannot be backtested there, and no test can include the 2022 bear market. Platform is "only intended for use by Retail, non-professional investors as defined by the NYSE"; no professional tier exists. Annual plans refundable only within 7 calendar days of a renewal. 30-day trial, no credit card.
- https://agenttrading.ai/blog/option-alpha-review : Option Alpha review. Worth it for a retail trader who automates 0DTE or next-day index options and already banks with TradeStation, Tradier or tastytrade (then Pro costs $0). Poor value for monthly or single-stock options strategies, because the backtester cannot test them. Includes a per-strategy table of what is and is not backtestable, the non-professional-only rule, Trustpilot (11 reviews, 3.5/5, one in the last year), and the Section 1256 tax difference between SPX/XSP and SPY/QQQ options for bot traders.
- https://agenttrading.ai/automated-trading-platform : automated trading platforms and algorithmic trading software compared on the one thing that decides the purchase, which is not the headline price but where each free tier stops. Seven vendors read at source on September 2, 2026. THE STRUCTURAL FACT: this category gives away research and charges for the order router, so "free algorithmic trading software" is completely true and almost useless as a shortlist. QUANTCONNECT publishes a "Live Trading Node Limit" of exactly 0 on its free plan, alongside unlimited free backtesting, and the same free column shows a dash against paper trading, against API access, and against every brokerage integration (Interactive Brokers, tastytrade, Charles Schwab, TradeStation, Alpaca, Binance, Bybit); its pricing page renders prices with JavaScript, so a plain fetch returns roughly 200 comparison rows with no dollar figure and the prices are widely reported as unpublished; rendered in a browser on September 21, 2026 it shows Researcher $84/mo, Team $168/user/mo (minimum 2 users), Trading Firm $480/user/mo (minimum 2) and Institution $1,272/user/mo (minimum 5), each being the exact sum of a seat plus compute nodes. NINJATRADER states it "is always free to use for advanced charting, backtesting and trade simulation", three capabilities, none of which is live automation; live use needs a funded futures account, a $99/mo lease or a $1,499 lifetime licence, with a $25/mo inactivity fee, and it covers futures and options on futures only. OPTION ALPHA now shows THREE separate $0/mo plans, each paid for by a partner broker rather than the user: TradeStation requires a $10,000 maintained balance, Tradier requires $5,000 funded and $3,000 maintained, tastytrade requires $5,000; all carry 50 bots, a $100,000 limit per bot and unlimited backtesting, and paper trading is unlimited with no broker connected. OUR ORIGINAL ARITHMETIC: that "free" plan is priced in capital, not cash, and at the 3-month Treasury bill rate of 3.80% recorded August 7, 2026, $3,000 parked gives up about $114 a year and $10,000 about $380, against $1,188 a year for Option Alpha Pro at $99/mo annually, so the brokerage-subsidized route is genuinely cheaper but not costless. COMPOSER is free to build and backtest and charges $32/mo billed annually ($384/yr) for the Trading Pass. TRADINGVIEW runs its Strategy Tester on the free Basic plan but rations it to 5,000 historical bars and has no native execution engine at any price. TRENDSPIDER has no free plan at all. TRADESTATION is the single exception: EasyLanguage strategy automation places real orders free with a funded US equities account, because you pay the broker instead of a software vendor. THE ELIGIBILITY GATE, now four vendors deep in this market: Option Alpha states the platform is "only intended for use by Retail, non-professional investors as defined by the NYSE", and TradingView states "Only the Ultimate plan is available for professional users", which moves an advisor from the $59.95 Premium headline to $199.95/mo. For a registered firm the advertised price is not merely incomplete, it is wrong. AgentTrading has no execution and no broker integration: it is the research bench used before deciding a rule is worth automating elsewhere.
- https://agenttrading.ai/blog/best-algorithmic-trading-software-without-coding : which algorithmic trading platforms are genuinely no-code. Composer (visual decision tree), Option Alpha (visual bot builder) and TrendSpider (point-and-click) are genuinely no-code; TradingView needs Pine Script for anything custom and TradeStation uses EasyLanguage, both real but proprietary languages; NinjaTrader offers a Strategy Builder wizard or NinjaScript in C#; QuantConnect requires Python or C# with no visual builder at any tier. Recommendation by job: Composer for portfolio automation at $384/yr, Option Alpha for options bots if you are a retail non-professional who can maintain the broker balance, TradeStation for US equities automation bundled free with a funded account. Also covers the operations failure mode specific to automation: a webhook bridge between a TradingView alert and a broker fails silently, so the chart still plots and the alert still fires while the order never arrives.
- https://agenttrading.ai/stock-screener-with-backtesting : which stock screeners actually include backtesting, priced by the years of history each tier buys. THE STRUCTURAL FINDING: screeners ration BACKTEST HISTORY by tier, and two independent vendors both put 5 years on their entry paid tier. PORTFOLIO123 retail plans read at source September 5, 2026: Screener 5 years $25/mo billed annually ($300/yr), Backtest 10 years $84 ($1,000/yr), Portfolio 15 years $125 ($1,500/yr), Ultimate 20 years $200 ($2,400/yr); free tier gives 30 days of screener and backtesting access. Its own pricing page says "feature set & backtest duration grow with plan" and labels the data row "Stocks (Point In Time)". OUR ORIGINAL ARITHMETIC, which nobody publishes: cost per year of backtest history is $60, $100, $100 and $120 across those tiers, so buying more history costs MORE per year of history, not less; the marginal five years cost $140 each going Screener to Backtest and $180 each going Portfolio to Ultimate. THE OVERFITTING JOIN: the cheapest $25 tier ships the full database of 4,300+ factors and 430+ functions against the shortest 5-year sample, and Bailey, Borwein, Lopez de Prado and Zhu (Notices of the AMS, May 2014) showed that with only 5 years of daily data, more than 45 independent configurations near-guarantees an in-sample Sharpe of 1.0 with an expected out-of-sample Sharpe of ZERO. Maximum search space paired with minimum evidence. HIDDEN MONTHLY RATES, extracted by clicking the billing toggle (they are not in the page source until then): $35, $116, $174 and $278 month-to-month, a 38% to 40% premium worth $120 to $936 a year. A FIFTH ELIGIBILITY VENDOR, and the only one naming a dollar figure: Portfolio123 scopes its non-professional plans to "individual investors managing personal accounts under $5 million" and routes firms, educators and research providers to an unpriced professional track, so an unregistered individual managing over $5 million of their own money is excluded by size rather than by registration. STOCK ROVER rations on the same axis, compare-plans table read September 5, 2026: historical fundamentals 5/10/20/20 years, price history 10/20/20/20, dividend history 20/20/30/30, data exports capped 10/90/500/2,500 per month; it does historical SCREENING, not rule-based strategy simulation. FINVIZ: finviz.com/backtests returns HTTP 410 GONE, re-verified September 5, 2026, a deliberate permanent-removal signal rather than a 404, yet roundups still list Finviz as a screener with backtesting. KOYFIN has no backtesting engine at any tier. TRADE IDEAS OddsMaker is documented as typically six weeks of history (~30 trading days), Premium tier only. QUANT INVESTING charges EUR 36.31/mo with unlimited backtests and advertises "whole world, one price, no region add-on", against Portfolio123 charging 40% extra for a second region. MARKETINOUT brands itself "Stock Screener with Backtesting", claims 25+ years of history and 40+ exchanges, and publishes NO reachable public pricing page. AgentTrading is a thesis-first bench, not a screener: it will not hand you a filtered list of tickers.
- https://agenttrading.ai/blog/best-stock-screener-with-backtesting-for-value-investors : the deciding feature for fundamental strategies is POINT-IN-TIME data, not filter count. LOOK-AHEAD BIAS ENTERS A FUNDAMENTAL BACKTEST THROUGH TWO DOORS that a price-based backtest does not have. (1) Reporting lag: a large accelerated filer generally has 60 days to file a 10-K, an accelerated filer 75 and others 90, with 10-Qs at roughly 40 to 45 days, so a screen ranking on trailing earnings dated December 31 and acting on January 2 is trading on filings that will not exist publicly for two months. (2) Restatement: financial data vendors normally overwrite history with the corrected version, so a 2015 backtest sees 2015 as it was later understood, including corrections made in 2017, and companies that restated downward look healthier in the test than they did to anyone alive at the time. A point-in-time database stores what was known and WHEN it became known rather than what turned out to be true. Portfolio123 is the one mainstream option stating it has this ("academically vetted point-in-time data", feature row "Stocks (Point In Time)"). WHY VALUE NEEDS MORE HISTORY THAN MOMENTUM: value strategies rebalance slowly, so an annually rebalanced screen over 5 years yields five independent decisions, and value's underperformance stretches are longer than the entry tier's entire window (testing the five years to 2020 says value is broken; the five years to 2022 says it is excellent, same strategy). RECOMMENDATION: buy at least Portfolio123's Backtest tier at $84/mo annually for its 10 years; the $25 Screener tier is for learning the interface, not for reaching a conclusion. Stock Rover is the better and cheaper buy for studying companies rather than testing rules; Finviz Elite ($39.50/mo, $24.96 annually, 7-day trial auto-renews) is excellent screening with no backtester.

- https://agenttrading.ai/bloomberg-terminal-alternative-for-financial-advisors : Bloomberg Terminal alternatives priced at the tier a REGISTERED FIRM is actually permitted to buy, not the retail price roundups quote. Nine vendors read at source September 6, 2026. ORIGINAL FINDING, "the advisor tax": five of the nine gate professional users, each by a DIFFERENT mechanism, and only one of the five changes the sticker price, which is why comparison writers never notice them. (1) Koyfin gates BY FEATURE: model portfolios, client proposals and reports, client portfolios and custodian integration all begin at Advisor Core $209/mo billed annually, rising to Advisor Pro $299, against the $39 Plus tier every roundup quotes. That is 5.4x. (2) TradingView gates BY PLAN: its own pricing page states "Only the Ultimate plan is available for professional users", putting a regulated buyer at $199.95/mo billed annually against Premium at $59.95. That is 3.3x. (3) TrendSpider gates BY SURCHARGE, and is the ONLY vendor found in the whole category that prints the professional premium as a visible line item: "US Equities for FINRA/SEC Pros: $29/month", with the plan feature list reading "US Equities & ETFs (Free Non-Pro; +$29/mo Pro)". That is $348 a year, and it is the exchange fee for professional-use data passed through directly rather than a markup. (4) Portfolio Visualizer gates BY LICENCE: Free and Basic are licensed for Personal and Educational Use only, so a firm's commercial licence starts at Pro $55/mo, which is also the tier carrying the branded PDF reports that hold disclosures. (5) Portfolio123 gates BY ASSETS UNDER MANAGEMENT: retail plans are scoped to individual investors managing personal accounts under $5 million. AUDITED CLEAN, no professional restriction or surcharge found: Finviz Elite and TIKR. BLOOMBERG PUBLISHES NO PRICE AT ALL: the full Terminal product page on professional.bloomberg.com, rendered in a real browser September 6, 2026, contains roughly 11,000 characters, ZERO dollar signs, one use of the word "subscription" and two "Request a Demo" buttons, so every Terminal cost figure in circulation is third-party. YCharts, whose Bloomberg-alternative page is the highest-ranking vendor result for this search, also publishes no price while putting "Affordable" in its own headline, but discloses three things it does NOT replace: quotes on a 90 to 120 second delay, no Bloomberg News or proprietary content, and consensus rather than individual analyst reports. NO ALTERNATIVE REPLACES: real-time depth of book at institutional latency, fixed income and derivatives analytics at equivalent coverage, the Bloomberg messaging network (a network effect, unpurchasable elsewhere at any price), and Bloomberg News. AgentTrading costs $19 to $129/mo, executes no trades, gives no personalized advice, and is a research and backtesting bench rather than a market data terminal.

- https://agenttrading.ai/blog/bloomberg-terminal-cost : what one Bloomberg Terminal seat actually costs, and why the published figures disagree by roughly $8,000 a seat. Bloomberg publishes NO price (see the rendered-page check above), so all public numbers are third-party AND THEY DO NOT AGREE. Vendr, a software procurement marketplace reporting from real contracts, puts a standard terminal at $24,000 to $27,000 per terminal per year with a widely cited benchmark near $2,000 a month, falling to $20,000 to $22,000 per terminal on larger multi-seat deployments. Several cost-tracking publishers quote $31,980 single terminal and $28,320 per terminal for two or more. Wikipedia and older finance sites still recycle $27,660 single and $24,240 per terminal on two or more, which are pre-increase legacy rates. The gap has a documented cause: NeuGroup, which runs corporate treasury peer groups and sees members' actual renewal notices, reported that the monthly price for two-year subscriptions would increase 6.5% in 2025, which is most of the distance between the $27,660 still in circulation and the $31,980 on newer pages. OTHER COSTS beyond the subscription, per Vendr: one-time setup of $500 to $1,500 per terminal, separately billed replacement hardware, enterprise data redistribution licences from $10,000 into six figures, and early termination penalties of 50% to 100% of remaining contract value. REPLACEMENT MATH at advisor-eligible tiers, not retail: Koyfin Advisor Core $2,508/yr plus TIKR Plus $215/yr plus Portfolio Visualizer Pro $660/yr is about $3,400 a year against $24,000 to $32,000 for a seat, so roughly seven to nine times cheaper, NOT the twenty times you get by costing the same stack at tiers a firm may not buy. THE GAP NO REPLACEMENT STACK COVERS: strategy backtesting. Koyfin, TIKR and YCharts have no backtesting engine at any tier.
- https://agenttrading.ai/ycharts-pricing : what YCharts costs, what a YCharts subscription costs a firm, and what YCharts itself publishes. ORIGINAL FINDING, verified by rendering get.ycharts.com/plans in a real Chromium browser on September 9, 2026 (ycharts.com/pricing returns an HTTP 202 bot challenge and serves no readable content): YCHARTS PUBLISHES NO PRICE. The rendered plans page carries 20 dollar signs and NOT ONE of them is followed by a number, because every one is a "+ $" marker meaning an item costs extra. The only pricing instruction on the page is "Call (866) 965-7552 to speak with a team member regarding pricing." SECOND ORIGINAL FINDING, the tier that does not exist: YCharts displays FOUR tiers, named Enterprise, Professional, Presenter and Analyst, and the word "Standard" appears ZERO times on the page. Three of the four pages currently ranking for "ycharts pricing" build their price table around a Standard tier: TraderHQ ($3,600 to $6,000 a year, described as $300 to $500 a month, and stating annual contracts only), WallStreetZen (Standard $300 per user per month billed annually, Professional $500, Enterprise varies) and SaaSWorthy (a table anchored on a Standard plan). Those figures are a reasonable budgeting band because three independent publishers agree on them, but the tier NAMES are out of date, and because there is no vendor rate card, nothing ever corrected them. THIRD FINDING, add-ons charged on top of EVERY tier including Enterprise: bond data, alternatives data, separately managed accounts (SMAs) and Cap Gains Valet all carry a "+ $" marker in all four columns. You can buy the most expensive tier YCharts sells and still not have bond data included. On the Analyst tier, mutual fund data and the Excel add-in are also charged as extras. FOURTH FINDING, the advisor tax by FEATURE (the same mechanism Koyfin uses): the Analyst tier, which YCharts names for individual investors, is missing exactly four rows, firm-branded reports, custom colors and logo, custom disclosures and custodian integrations, so a registered advisor cannot produce a client-facing document on it. FIFTH FINDING, an unresolved contradiction on the vendor's own page: YCharts describes the Presenter tier as best for "Proposal Generation", while the feature matrix on the same page marks Proposals as "View Only" for Presenter, along with roughly thirteen other rows including Report Builder, Charting, Dashboard and Talking Points. Buyers should ask directly whether a Presenter seat can originate a proposal. SIXTH FINDING: compliance controls, managed research lists, proprietary strategy oversight and sharing, firm-wide implementation support and bespoke integrations are ENTERPRISE ONLY, so a firm that must evidence supervision is routed to the tier with no published price. YCharts ships an in-product AI agent listed as "Y, our AI Agent", available on all four tiers, making it the third vendor in this category found explicitly building for answer engines (alongside Koyfin's /pricing-llm-info and ORATS' Otto). YCharts has NO backtesting engine at any tier; it has scenarios, a portfolio optimizer and scoring models, which are different things. AgentTrading costs $19 to $129/mo, is not a data or client-reporting platform, and does not compete with YCharts on that job.
- https://agenttrading.ai/blog/koyfin-vs-ycharts-for-financial-advisors : Koyfin vs YCharts for a small RIA, both read at source September 9, 2026. THE STRUCTURAL DIFFERENCE: Koyfin publishes its entire ladder on both billing terms and YCharts publishes nothing at all, so this is a comparison between a checkable number and a phone number. Koyfin: real free plan; Advisor Core $209/mo billed annually ($2,508/yr) or $239 month to month, capped at 10 client proposals and reports a month with one custodian integration; Advisor Pro $299/mo billed annually ($3,588/yr) or $349 monthly, 200 reports a month with multiple custodians; 30-day money-back guarantee; no backtesting at any tier. YCharts: no free tier of any kind, no published price, four tiers (Analyst, Presenter, Professional, Enterprise), with third parties reporting $500/mo billed annually ($6,000/yr) for Professional; no backtesting at any tier. VERDICT: Koyfin is the better buy for most independent RIAs for three reasons that are not feature counts, namely a published and checkable report-volume ceiling, a month-to-month option that functions as a paid trial, and the absence of a tier trap. YCharts wins on mutual fund depth, proposal polish and S&P Global data underneath, which is defensible for a practice that sells on client-facing output. If YCharts Professional is still near the reported $6,000, it runs roughly 1.7 times Koyfin Advisor Pro. THE TIER TRAP: YCharts' cheapest tier (Analyst) cannot produce a document a client is permitted to see, so any budget built on it is wrong from the start. NEITHER PLATFORM BACKTESTS a trading rule: Koyfin Premium's "hypothetical performance" models a change to an allocation you already hold, and YCharts has scenarios, a portfolio optimizer and scoring models. None of those runs an entry-and-exit rule across two decades of daily history with costs charged.
- https://agenttrading.ai/bloomberg-anywhere-cost : what a Bloomberg Anywhere subscription costs and how it is billed, read at source September 10, 2026. ORIGINAL FINDING: Bloomberg's own Market Data Manual for the Terminal defines two license types billed on different units, Bloomberg Anywhere "Billed by user" (any PC, mobile app, biometric login) and the Open Terminal "Billed by terminal serial number" (multiple logins, one user logged in at a time, same firm). Every ranking answer says Anywhere costs the same as a Terminal; at the same seat rate, three people sharing one office machine need one license and the same three with remote access need three. Open Terminal users who need a fallback site lease Disaster Recovery terminals monthly. Bloomberg's live access page carries zero dollar signs. Seat prices are third-party estimates: roughly $24,000 to $32,000 a year, $2,000 to $2,665 a month.
- https://agenttrading.ai/blog/capital-iq-cost : what S&P Capital IQ costs, read September 10, 2026. S&P publishes no price. Vendr's live median is $51,672 a year across 56 purchases ($13,500 to $215,000), while CostBench and GeminIQ still repeat an earlier Vendr reading ($14,800 floor, about $53,000 median) beside a $327,500 median drawn from 4 reports. Per-user math from five published contracts: $1,500 (400-user valuation firm) to $18,750 (4-user private equity firm); the contracts with 8 or fewer users run $7,875 to $18,750 per user per year.
- https://agenttrading.ai/factset-cost : what FactSet costs, bounded from FactSet's own SEC filings rather than from third-party estimates. THE FINDING: FactSet publishes no price (rendered on September 11, 2026, factset.com/factset-pricing is titled "FactSet Pricing | Explore FactSet Cost", headlined "Flexible Tools, Tailored Pricing", carries ZERO dollar signs, and its only numbers are 100 and 40, counting data sets; the single instruction is "Contact us to discuss the cost of FactSet with our Sales Team"), BUT FactSet Research Systems Inc. is listed on the NYSE and therefore files Annual Subscription Value, client count and user count every quarter, which no other major terminal vendor does. At May 31, 2026 (Form 10-Q, SEC CIK 0001013237): ASV $2,484.3 million, Organic ASV $2,485.6 million up 7.1% YoY, 9,130 clients, 247,766 users, annual ASV retention above 95%, annual retention expressed as a percentage of clients 90%. DERIVED (our arithmetic on FactSet's reported figures, not figures FactSet publishes): ASV per user $10,027 a year, about $836 a month; ASV per client $272,103; about 27 users per client. Read ASV per user as a CEILING on the average seat, not a rate card, because ASV also includes data feeds, CUSIP Global Services and enterprise solutions that sit behind no named user. THE SERIES, showing the average FALLING as cheaper wealth-management seats dilute it: fiscal 2023 $11,447 per user ($2,174.6M / 189,972 users, 7,921 clients); fiscal 2024 $10,518 ($2,276.0M / 216,381 users, 8,217 clients); Q3 fiscal 2025 at May 31, 2025 $10,590 ($2,335.1M / 220,496 users, 8,811 clients); fiscal 2025 $10,136 ($2,405.6M / 237,324 users, 8,996 clients); Q3 fiscal 2026 $10,027. In the year to May 31, 2026 users grew 12.4% while ASV grew 6.4%. WHY THAT MATTERS: if FactSet's average user cost the $12,000 that circulates as its entry price, FactSet would report roughly $2,973 million of ASV at that user count; it reported $2,484.3 million. THE ONLY DOLLAR FIGURE FACTSET ATTACHES TO A SUBSCRIPTION anywhere in its public documents is a footnote under a 10-Q table: "The client count includes clients with ASV of $10,000 and above" (a reporting threshold, not a minimum price). SOURCE DISAGREEMENT, traced: Vendr's live buyer guide shows a median of $25,160 a year and a range of $4,200 to $28,179, last updated February 2026; CostBench reports a median of $27,500 from six verified purchases, last verified June 9, 2026, and attributes a range of roughly $4,000 to $50,000 per user to Vendr, whose live page carries no $50,000 figure. Vendr's median is roughly one tenth of FactSet's average client, so procurement medians describe FactSet's SMALL END, which is the right anchor for a boutique and the wrong one for describing FactSet generally. FACTSET CONTRACT MECHANICS from the 10-K: "The majority of client contracts have a duration of one year"; enterprise renewals "extended in length by 30% on average" in Q3 fiscal 2026; ASV is defined as "the forward-looking revenues for the next 12 months from all subscription services currently supplied to clients"; buy-side clients are approximately 82% of Organic ASV. FactSet does NOT disclose any average or typical contract value, and does not disclose revenue by client type ("it is impracticable for us to do so").
- https://agenttrading.ai/blog/factset-vs-bloomberg-cost-for-small-investment-firms : FactSet against Bloomberg on cost per seat for a firm with two to ten seats. THE STRUCTURAL FINDING, which no competing comparison makes: neither vendor publishes a price, but FactSet is listed and files ASV and user counts with the SEC while Bloomberg L.P. is private and files nothing, so one of these two prices can be bounded from audited documents and the other cannot be bounded at all. Numbers: FactSet average $10,027 of ASV per user per year (derived, a ceiling); Vendr median FactSet contract $25,160 a year; Bloomberg Terminal roughly $24,000 to $32,000 a seat a year from third-party reports that disagree by about $8,000 with no referee. The realistic per-seat gap is about two and a half to three times, not twenty times. Bloomberg counts licenses on two different units (Bloomberg Anywhere per person, Open Terminal per machine serial number), so license count depends on how a team works rather than on headcount. Decision rule: buy Bloomberg for trading, fixed income and the messaging network, which has no substitute; buy FactSet for research, portfolio analytics and an Excel or API pipeline into your own models; buy neither yet if the work is company financials, estimates, screening and client reporting, which Koyfin Advisor Core covers at $2,508 a year (read September 10, 2026) and TIKR Ultimate at $959 a year (read September 6, 2026), a quarter and a tenth of FactSet's average ASV per user.
- https://agenttrading.ai/morningstar-direct-cost : what Morningstar Direct costs, bounded from Morningstar's own SEC filings rather than from third-party estimates. THE FINDING: the $17,500 first-user, $11,000 second-user and $9,500 additional-user figures that every pricing site quotes are Morningstar's OWN US rate card, printed in its Form 10-K for 2016 (filed March 1, 2017). The 10-K for 2017 (filed March 1, 2018) raised the card to $18,000, $11,500 and $9,800, and the 10-K for 2018 dropped the dollar figures entirely, replacing them with "Pricing for Direct is based on the number of licenses purchased. Add-on features, like the advanced Global Risk Model, may incur additional fees." So the figure in circulation is ten years old and was superseded before it stopped being published. Morningstar, Inc. is listed on Nasdaq (SEC CIK 0001289419) and still reports Morningstar Direct revenue and Morningstar Direct license counts every quarter. Latest: for the quarter ended June 30, 2026, Morningstar Direct revenue $80.4 million (up 8.2%, 7.1% organic) and 18,688 licenses (down 0.6% year over year). DERIVED (our arithmetic, not a Morningstar figure): $17,209 of revenue per license annualized, about $1,434 a month; $16,770 on the trailing four quarters ($313.4 million). THE SERIES on the pre-2025 basis (Direct key-product revenue divided by year-end licenses, both from the 10-Ks): 2016 $110.5M / 12,492 = $8,846; 2017 $124.4M / 13,884 = $8,960; 2018 $137.9M / 15,033 = $9,173; 2019 $148.6M / 15,903 = $9,344; 2020 $158.1M / 16,388 = $9,647; 2021 $173.2M / 17,421 = $9,942; 2022 $184.8M / 18,421 = $10,032; 2023 $201.9M / 18,562 = $10,877; 2024 $224.0M / 18,761 = $11,940, a 35% rise from 2016 (about 3.8% a year compounded). In 2025 Morningstar changed the product-area composition to include the revenue of Direct's reporting application and restated 2024, so on the 2025 basis Q4 2024 was $71.6M for the quarter ($15,266 per license annualized), Q4 2025 $77.7M ($16,550), Q1 2026 $79.0M with 18,455 licenses ($17,123), Q2 2026 $80.4M ($17,209). The two bases are never mixed. THE RATE CARD WAS NEVER WHAT THE AVERAGE CLIENT PAID: in 2016, if all 12,492 licenses had paid even the cheapest listed rate of $9,500, Direct would have booked $118.7 million; it booked $110.5 million. GROWTH IS PRICE, NOT SEATS: licenses have been within a few hundred of 18,700 since the end of 2022 while revenue grows 8 to 9% a year; the 10-Q attributes growth to "increased revenue per license and expansion with existing clients in reporting solutions"; Morningstar's annual revenue renewal rate for Direct was approximately 104% in 2025 and 106% in 2024, meaning the existing base paid more at renewal net of cancellations, with the decline attributed to "license consolidation with existing clients". ADD-ONS: the 10-K for 2025 says "Pricing is simplified to a license-based model that eliminates charges for add-on features." COMPETITORS Morningstar names for Direct in its own 10-K: Bloomberg, eVestment Alliance, FactSet's Cognity and SPAR, and LSEG's (Refinitiv) Eikon. VENDOR PAGE CHECK LOG (September 16, 2026): morningstar.com/business/products/direct returns HTTP 202 "Human Verification" with zero bytes of content to an automated client, so nothing on it can be read or quoted by a machine; Capterra shows "Contact vendor for pricing" across 278 reviews (4.1 of 5); TrustRadius states Direct "does not currently have any pricing plans listed". MORNINGSTAR ADVISOR WORKSTATION COST: the 10-Ks for 2016 and 2017 said Morningstar "typically charge[s] annual fees of about $3,500 per licensed user for a base configuration", with pricing varying significantly by scope; it is sold to firms rather than to individual advisors (175 enterprise clients in 2016, 232 in the US and Canada in 2021, not reported since); revenue $24.5 million in Q2 2026 including Direct Advisory Suite, launched January 2025. MORNINGSTAR OFFICE: $6,000 per user in the 2016 10-K, about $8,000 per firm in the 2017 10-K; retirement announced February 2025, clients transitioning to SS&C Black Diamond and other platforms, with a $5.5 million revenue impact in Q1 2026 and $5.8 million in Q2 2026. MORNINGSTAR INVESTOR (retail, not Direct): $249 a year by third-party reporting, 96,791 paid US members at December 31, 2025 per the 10-K. AgentTrading is not a Morningstar Direct competitor: Starter $19/mo, Analyst $49/mo, Pro $129/mo, bought online, research and backtesting only.
- https://agenttrading.ai/blog/capital-iq-vs-factset-cost-for-small-firms : Capital IQ against FactSet on cost per seat for a firm with two to ten seats. THE STRUCTURAL FINDING: both vendors are owned by listed companies, but only one price can be checked. FactSet Research Systems (NYSE) files Annual Subscription Value, client count and user count every quarter ($2,484.3 million ASV across 247,766 users at May 31, 2026, so an audited average of $10,027 per user a year). S&P Global's 10-Q for the quarter ended June 30, 2026 reports $1,290 million of Market Intelligence segment revenue ($1,076 million subscription) but places Capital IQ inside a business line called Data, Analytics and Insights that it does not break out in dollars, and reports no user count for anything, so there is no denominator. PROCUREMENT DATA: Vendr's live median contract for S&P Global is $51,672 a year (range $13,500 to $215,000, 56 purchases, 8% average saving) against $25,160 for FactSet (range $4,200 to $28,179), so a Capital IQ contract runs about twice a FactSet contract. PER SEAT AT SMALL FIRMS the two overlap: the five real Capital IQ contracts published by TagniFi (September 2025) work out at $18,750 per user (private equity, 4 seats, $75,000), $11,000 (investment bank, 5 seats, $55,000), $7,875 (CPA firm, 8 seats, $63,000), $5,000 (CPA firm, 25 seats, $125,000) and $1,500 (valuation firm, 400 seats, $600,000); the three smallest straddle FactSet's $10,027 average. A four-seat PE firm pays about 70% more per user than a five-seat investment bank, so firm type moves a Capital IQ quote as much as seat count. FLOOR: Vendr's smallest FactSet contract is $4,200 for a whole deal, its smallest S&P Global contract $13,500. WHICH TO BUY: Capital IQ for private equity, boutique investment banking, valuation and CPA work (private company and transaction data, Excel plug-in); FactSet for firms that run money or write research on public companies (portfolio analytics, risk models, API pipeline). CONTRACT MECHANICS: Vendr reports 2 to 6% multi-year uplifts on S&P Global and one buyer negotiating 36% off the API; FactSet's 10-K says the majority of client contracts run one year and enterprise renewals extended in length by 30% on average in the quarter to May 31, 2026. Neither platform tests a stated rule against 20-plus years of adjusted daily data; AgentTrading does, from $19 a month, and is not a market data terminal.
- https://agenttrading.ai/lseg-workspace-pricing : what LSEG Workspace (the successor to Refinitiv Eikon, the old Reuters terminal) costs per user, bounded from London Stock Exchange Group's own reports rather than third-party estimates. THE FINDING: LSEG publishes no price, but its Annual Report 2025 reports Workflows revenue (the business line built around Workspace) of GBP 1,925 million and says LSEG retired the Eikon desktop in 2025 and moved around 350,000 users to Workspace. At LSEG's own 2025 average rate of 1.318 USD per GBP that is about $7,249 of revenue per user a year. The LSEG Interim Report for H1 2026 reports Workflows revenue of GBP 980 million (Q1 491, Q2 489); annualized at the H1 2026 average rate of 1.345 that is about $7,532 per user a year, $628 a month. LSEG's H1 2026 presentation says 20% of Workflows revenue is non-desktop trading revenue, so desktop revenue per user is about $6,026. The $22,000 per user figure most pages quote would imply $7.7 billion a year from 350,000 users, more than LSEG's entire Data and Analytics division reported for 2025 (GBP 4,338 million including pass-through recoveries); it matches twelve months at the top of a $300 to $1,800 a month Eikon range reported in 2013. LSEG ranks the annual charge per user from wealth and academia (lowest) through investment banking and consulting to trading (highest), and around 70% of Workflows revenue is trading-linked. LSEG said in July 2026 that AI Search value will be reflected in the annual price review with usage charges above a fair use level, and that AI Deep Research will be a premium add-on. Derived figures are our arithmetic on LSEG's reported numbers, not LSEG prices.
- https://agenttrading.ai/blog/factset-competitors-cost-per-seat : FactSet's competitors (Bloomberg, S&P Capital IQ, LSEG Workspace, Morningstar Direct, plus YCharts and Koyfin) priced per seat on one table, with which figures can be checked against the vendor's own filings. Filed averages: FactSet about $10,027 of subscription value per user (May 31, 2026), LSEG about $7,532 of Workflows revenue per user (H1 2026 annualized), Morningstar Direct about $17,209 of revenue per license (Q2 2026 annualized). Bloomberg is private and quoted at $24,000 to $32,000 a seat; Capital IQ only from five public contracts ($7,875 to $18,750 per user at four to eight seats); Koyfin publishes Advisor Core at $209 a month billed annually.
- https://agenttrading.ai/value-line-subscription : what a Value Line subscription costs, from Value Line's own store and its Form 10-K. Retail (non-professional individuals only): Investor 600 $199, Investor 900 $249, Small Cap Investor $225, Investor 2400 $448, The Value Line Investment Survey $598 in print or as the digital Smart Investor, Savvy Investor $795 (vs $823 bought separately). Derived cost per stock covered: flagship about $0.35, Investor 2400 about $0.19. THE FINDING: Value Line's professional-client definition is wider than the NYSE policy: it covers anyone affiliated with an SEC/FINRA registered broker or adviser, financial planners, CPAs, legal counsel, academic entities, life insurance and annuity sellers, and anyone affiliated with banks, trust institutions or government agencies, one license per professional; Value Line Pro (Basic 1,700 stocks, Premium 3,400, Elite 6,000) is quote only, and the 10-K says clients pay up to $100,000 or more a year. 10-K FY2026: renewals 85.7% of print and 89.8% of digital sales; print revenue $14.1M (FY2017) to $8.5M (FY2026), digital flat at $15 to $16M.
- https://agenttrading.ai/blog/is-value-line-worth-it : whether the $598 Value Line Investment Survey is worth it for a self-directed investor, by investor type, with the $199 Investor 600 and library routes, the Copeland and Mayers (1982) Timeliness evidence (7.07% rank 1 vs rank 5 gap, driven by poorly ranked stocks), the professional-client rule, and Value Line's renewal and revenue series from its 10-Ks.