Backtesting software: the best backtesting platforms and trading strategy testers
Type the rule the way you would say it. AgentTrading restates your idea as an explicit entry-and-exit rule card before anything runs, tests it on 20+ years of adjusted daily data, and stamps the verdict, even when the rule loses.
- 1 THESIS
- 2 EVIDENCE
- 3 BACKTEST
- 4 RISK
- 5 VERDICT
02 EVIDENCE · FUNDAMENTALS
04 RISK · IN PLAIN ENGLISH
Sample scenarios, not a live backtest of what you typed. Past performance does not guarantee future results. Educational analysis only, not financial advice.
In short
Backtesting software applies a trading rule to historical market data and reports how the rule would have behaved, and a strategy tester is only as good as the rule you can actually read before it runs. AgentTrading restates your plain-English idea as an explicit rule card, such as "Entry: 50-day crosses above the 200-day. Exit: cross below", before anything executes, so you always know exactly what is being tested. The rule then runs on 20+ years of split- and dividend-adjusted daily data with a 0.1% cost per trade assumed by default, and every assumption is printed on the result: date range, parameters, dividend treatment, cost. The verdict is stamped even when it hurts. The classic AAPL golden cross, tested honestly in our illustration, UNDERPERFORMED buy-and-hold over long stretches, and AgentTrading shows that instead of burying it. Each result carries the trade count with a small-sample warning, because a rule that traded nine times in twenty years proves close to nothing. Most backtesting software makes you meet it halfway: TrendSpider ($59 to $349 per month) expects indicator fluency, Portfolio Visualizer is a long configuration form with no explanation of the result, and Tickeron ($60 to $250 per month) leads with advertised win rates, the exact habit we avoid. AgentTrading starts at $19 per month with no code, no parameter sweeps, and no execution. It is educational analysis only, and past performance does not guarantee future results.
Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.
SIDE BY SIDE - PRICING CHECKED AUGUST 2026
The eleven platforms that genuinely backtest, side by side
List prices read off each vendor's own pricing page: most on August 4, 2026, NinjaTrader on August 11, and TrendSpider and TradeStation re-read on August 16 because both were being misreported elsewhere. TrendSpider's AI assistant turns out to be metered and billed on top of the plan, and TradeStation's widely quoted $99.99 and $59.95 fees apply to software-only customers rather than to people with a funded account. Read the limitation column twice, because it is the part no vendor puts on a landing page.
| Tool | Best for | Price (August 2026) | The honest limitation |
|---|---|---|---|
| AgentTrading | One rule, stated in plain English | From $19/mo | No execution, no parameter sweeps, no option chains, no real-time scanning. |
| QuantConnect | Coders who want a real research stack | Free tier, unlimited backtests | No code, no backtest. Paid compute is configured per seat and node. |
| AmiBroker | Fast local testing you own outright | $299 to $499 one-time | Windows only. AFL code required. Data feed not included in the price. |
| Portfolio Visualizer | Asset allocation and factor tests | Free tier; $30 or $55/mo annually | Portfolio level, not single-ticker rules. Statistics with no interpretation. |
| TradingView | Chart-native testing via Pine Script | Free; $14.95 to $239.95/mo | Real tests need Pine Script, and a borrowed script can quietly cheat. |
| TrendSpider | No-code chart-condition strategies | $59 to $349/mo, AI billed separately | Backtest depth is rationed by tier, and the Sidekick AI is metered on top of the plan. |
| TradeStation | Broker-hosted testing you can automate | Free with a funded account | EasyLanguage for anything complex, and the desktop platform is the only place the full suite runs. |
| NinjaTrader | Futures traders who want a real optimization suite | Free plan, $99/mo, or $1,499 lifetime | Futures only, no cash stocks or ETFs. NinjaScript is C#, and it is Windows first. |
| Fidelity Strategy Testing | Free technical tests inside your broker | Free with an account | Technical equity rules only. No fundamentals logic, no options backtest. |
| thinkorswim | Reviewing history by hand | Free with a Schwab account | thinkBack and OnDemand are manual. Automation needs thinkScript. |
| Trade Ideas | Real-time scanning, with testing on Premium | $127 to $254/mo | Backtesting sits on Premium only, and Holly's reasoning stays closed. |
Prices are the vendors' public list prices, checked August 2026, and vendors change them. Verify current pricing and capabilities with each vendor before you buy.
AgentTrading
One rule, stated in plain English
You type the thesis the way you would say it out loud, such as "hold NVDA above its 50-day moving average", and the bench restates it as an explicit rule, shows you that rule before it runs, tests it on 20+ years of adjusted daily data with costs charged, reads the drawdown in plain English, and stamps an honest verdict including UNDERPERFORMED when the idea loses to simply holding. Where we lose: we do not sweep thousands of parameter combinations looking for the best version of your rule, we hold no chain-level option quotes, we do not scan the market live, and we will never place a trade. If you want a machine to search for a strategy, this is deliberately the wrong tool.
QuantConnect
Coders who want a real research stack
The most capable platform on this list by a wide margin, and the closest thing to a professional quant environment available to an individual. The free tier includes equities, indexes, forex, crypto, futures, and options data with unlimited backtesting, and you write strategies in Python or C# against a documented engine that models order fills rather than assuming them. Paid tiers are not fixed prices: you configure organization seats, backtesting and live-trading compute nodes, datasets, and support, then see a total. The barrier is real programming ability. There is no builder to fall back on, and a beginner will spend weeks before the first trustworthy result.
AmiBroker
Fast local testing you own outright
The value pick for anyone who tests seriously and dislikes subscriptions: Standard is $299, Professional $379, and Ultimate Pack Pro $499, each with 24 months of free upgrades, after which upgrades run at half the new-license price. It is extremely fast, runs entirely on your own machine, and its AFL language handles portfolio-level tests, optimization, and walk-forward analysis that most no-code tools cannot touch. Three catches: it is Windows-only, AFL is a genuine programming language with its own idioms, and you must supply market data separately, so the true first-year cost is higher than the sticker.
Portfolio Visualizer
Asset allocation and factor tests
The default for portfolio-level work: allocation backtests, factor regressions, Monte Carlo, and rolling returns. The free tier analyzes up to 15 assets, including backtesting and Monte Carlo, but saves nothing and exports nothing; Basic is $30 per month billed annually ($360 a year, 50 saved models) and Pro is $55 ($660 a year, 150 models plus custom tax analysis), read September 4, 2026. Only the Pro tier is licensed for commercial use, so an advisory firm cannot rely on the $30 figure. If you want to compare a 60/40 against a three-fund portfolio, nothing here beats it. It is also a long configuration form, and the output is a table of statistics rather than a plain-English read of what the risk actually felt like.
TradingView
Chart-native testing via Pine Script
The Strategy Tester attached to the best charting platform in retail, and the free tier can run it, which is why it is the most common first backtest anyone performs. Paid plans run $14.95, $34.95, $69.95, and $239.95 per month, less billed annually. Two honest cautions. First, anything beyond a stock indicator strategy requires writing Pine Script, so the no-code promise runs out quickly. Second, the community library is full of scripts with lookahead bias baked in, and a strategy that repaints its signals will show a beautiful equity curve that was never achievable. Check the source before trusting the result.
TrendSpider
No-code chart-condition strategies
The strongest no-code condition builder for technical strategies, paired with automated trendlines, multi-timeframe analysis, and raindrop charts. Backtesting is included on every tier, but the depth is rationed: Standard is capped at 2-hour bars, 2,000 bars of history, and 50 symbols per test, rising to 1-minute bars, 30,000 bars, and 510 symbols on Advanced, with a $39 per month add-on to lift Standard and Premium to 20,000 bars. Re-read on August 30, 2026, and TrendSpider has repriced upward since our August 16 check: Standard is now $89 month-to-month or $59 billed annually, Premium $149 or $99, Enhanced $199 or $133, Advanced $349 or $233, Business from $399, with paid 14-day trials at $19 to $49. The page also runs an introductory offer showing Standard at $52.38 against a struck-through $82, and the terms attached to it state that it is available to new customers only and that the discount applies to the first invoice only, so the number in the largest type is not the rate you will pay in month two. The part reviews miss is that Sidekick, the AI assistant, is a separate subscription. Every plan includes 25 AI messages a month, and more costs $49, $129, or $349 per month on top of the base plan.
TradeStation
Broker-hosted testing you can automate
One of the few brokers with a genuine historical backtester rather than a replay tool, and the pricing is widely misreported. Re-read on August 16, 2026: there is 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 people who want the software without a brokerage account. Testing is serious: decades of history, optimization, rolling walk-forward analysis, intrabar order generation, and modeled slippage and commissions, with Portfolio Maestro running standard, optimization, and walk-forward tests plus Monte Carlo across a basket. Strategy Builder covers simple rules without code; anything complex means EasyLanguage. Because it is also a broker, a validated strategy can be automated in the same place, which is exactly the moment to be careful.
NinjaTrader
Futures traders who want a real optimization suite
The deepest backtesting environment a retail trader can reach without paying for it. The Strategy Analyzer runs exhaustive and genetic parameter optimization, rolling and anchored walk-forward tests, and Monte Carlo, and Market Replay re-runs a past session tick by tick so you can trade it by hand. The free plan includes all of that, with paid tiers buying cheaper commissions rather than more software: $1.29 per side on standard contracts and $0.39 on micros for free users, falling to $0.59 and $0.09 on the $1,499 lifetime license (verified August 2026). Watch the $25 monthly inactivity fee. The catch is scope. NinjaTrader is a futures broker, so there are no cash equities, ETFs, or stock options to test, and the CME single stock futures it added in July 2026 are contracts on a company rather than shares. Custom strategies mean C#, and Mac users run a virtual machine.
Fidelity Strategy Testing
Free technical tests inside your broker
The most overlooked free option in this list, and the first thing to check if you already bank with Fidelity. Its web-based Strategy Testing tool offers prebuilt and customizable technical templates, runs single-symbol and multi-symbol tests, and models dividends, interest, and commissions, which is more rigor than several paid tools provide. Two boundaries worth knowing: the templates are technical only, so a rule involving earnings growth or valuation is out of scope, and its options tooling is forward-looking probability analysis rather than a historical options backtest. Active Trader Pro, despite the name, has no backtesting engine.
thinkorswim
Reviewing history by hand
Free with a Schwab brokerage account, with no platform fee, no data fee, and no minimum, which makes it the cheapest serious charting package anywhere. For history it gives you three tools: thinkBack for reviewing historical options prices, OnDemand for replaying the market against a virtual $100,000 paper account across roughly 14 to 15 years, and strategy reports on charts. Note what is missing: OnDemand is a manual replay, not a batch test, so you cannot ask it to evaluate one rule across two decades and hand you a summary. Automated custom backtesting means learning thinkScript.
Trade Ideas
Real-time scanning, with testing on Premium
Primarily a real-time scanner rather than a backtesting platform, included here because it is so often shortlisted alongside them. Check the billing cycle carefully: the $89 and $178 figures quoted across most review sites are annual-plan rates, and month-to-month is $127 for TI Standard and $254 for TI Premium. Strategy backtesting and Holly, its AI engine, sit on Premium, not Standard, which materially changes the real cost of using it for testing. The deeper limitation is direction. It pushes machine-generated setups at you and the logic inside Holly stays opaque, so you cannot bring your own thesis and audit the evidence behind it.
Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.
WHAT YOU GET - BACKTESTING SOFTWARE
Backtesting software, run on the bench
Rules you can read, tests you can trust
The tester never runs anything you have not seen. The restated rule card, entry and exit in plain language, is the contract for the whole test.
The losing case stays on the menu
The AAPL golden cross scenario UNDERPERFORMED buy-and-hold in our 20-year illustration, and it ships as a headline example. That is the honesty you want in a tester.
Small samples get called out
A rule that traded nine times in twenty years proves very little. The tester counts trades and says so in plain English before you over-read the chart.
Costs included by default
Every test assumes 0.1% per trade, so a rule that only works with zero costs fails here first. Pro users can set their own cost and slippage assumptions.
Every assumption printed on the result
Date range, cost per trade, rule parameters, and dividend treatment appear with every chart. If an assumption would change the answer, you can see it rather than guess at it.
Minutes instead of a spreadsheet weekend
The manual version is a data download, a formula sheet, and an evening gone: roughly 4 to 6 hours per idea. The bench runs the same test in minutes and logs what it assumed.
HOW IT WORKS - 4 STEPS
From a sentence to a stamped verdict
Type the rule
"Buy the 50/200 golden cross on AAPL, sell the cross down." Plain English is the input format.
Confirm the rule card
Entry and exit restated explicitly before the run. If it is not the rule you meant, rephrase and it updates.
Run 20+ years of history
Split- and dividend-adjusted daily data, dividends reinvested, 0.1% cost per trade by default, and the worst drawdown window shaded on the chart.
Read the verdict honestly
HELD UP, MIXED, or UNDERPERFORMED against buy-and-hold, with trade count and a sample-size warning. Historical, not predictive.
Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.
On the same bench
The tester pairs with the trading strategy builder, which turns your sentence into the rule card the test runs on, and the historical stock data page explains the adjusted 20-year record every test runs on. Test one ticker on stock backtesting, a whole basket on portfolio backtesting, and for the strategy families people test most see momentum backtesting for trend rules, moving average crossover backtesting for golden cross and 50/200 rules, and dividend backtesting for total-return income rules. If you are still choosing between platforms, the table above compares eleven of them on re-verified list prices, TrendSpider pricing breaks down the tier that actually carries the backtest depth you need, and if the goal is to automate the rule once it survives, automated trading platforms compares what each vendor charges to actually send an order, and the honest reads on the two most common paid alternatives are the TrendSpider alternative and the TradingView alternative, which covers testing a rule without writing Pine Script. If your broker might already include a tester, does TradeStation have backtesting answers the one most people get wrong. Swing traders sizing setups should see AI swing trading tools and swing trading tools, and the Tickeron alternative explains why we never advertise win rates. Plans start at $19 a month.
QUESTIONS - ASKED AND ANSWERED
Backtesting software: the common questions
What is a trading strategy tester?
A trading strategy tester takes a rule with a defined entry and exit and shows how it performed on historical data. AgentTrading restates your idea as an explicit rule card before anything runs, tests it on 20+ years of adjusted daily data, and stamps an honest verdict, even when the rule loses.
How do you test a trading strategy?
Define the entry and exit, run the rule on clean historical prices, and compare it against buy-and-hold with trading costs included. In AgentTrading you type the rule in plain English, confirm the restated rule card, and it runs on 20+ years of split- and dividend-adjusted data at 0.1% cost per trade.
Can you test a strategy without coding?
Yes. You type the rule as a sentence, such as "buy the 50/200 golden cross on AAPL, sell the cross down", and AgentTrading restates it as an explicit entry-and-exit card you confirm before the test. No Pine Script, no indicator fluency, no spreadsheet. Plans start at $19 a month.
How accurate is strategy testing?
A test is only as good as the rule and the data behind it. AgentTrading uses 20+ years of split- and dividend-adjusted daily data, assumes 0.1% cost per trade, and flags small samples: a rule that traded nine times proves little. It shows what happened in history, never what will happen next.
What is backtesting software?
Backtesting software applies a trading rule to historical market data and reports how the rule would have behaved, so an idea can be judged on evidence rather than intuition. A useful backtester states the rule explicitly before running, uses split- and dividend-adjusted prices, charges realistic trading costs, and compares the result against simply buying and holding the same asset over identical dates.
Is backtesting actually worth it?
Yes, but for the opposite reason most people expect. Backtesting rarely finds a winning strategy; what it does reliably is kill bad ones cheaply. Learning in ten minutes that a rule underperformed buy and hold across 20 years, or that it required sitting through a 55% drawdown, saves real money and years of drift.
How much historical data do you need for a backtest?
At least 20 years. A shorter window can miss every real bear market and produce a strategy that looks robust only because it was never tested. Twenty-plus years of adjusted daily data forces a rule through the dot-com unwind, 2008, and 2020, which is the minimum bar for taking a result seriously.
Why do backtests look better than live results?
Four reasons, and they compound: overfitting to historical noise, trading costs left out or understated, look-ahead bias from data the strategy could not have had, and survivorship bias in the ticker universe. A backtest that prints every assumption on the result is the only kind you can audit for these.
What is the best backtesting software?
There is no single winner, because the answer turns on whether you write code. QuantConnect is strongest for coders and has a free tier with unlimited backtests. Portfolio Visualizer leads for asset allocation. TrendSpider leads for chart conditions. AgentTrading, from $19 per month, is built for testing one rule you can state in a sentence, with no code and no builder to learn.
Which backtesting software is best for beginners?
Start with something free that matches your question, because a beginner rarely knows the shape of the question yet. Portfolio Visualizer handles allocations free for up to 15 assets, and your broker may already include a tester, as Fidelity and TradeStation do. Avoid tools that need Pine Script, AFL, or C# until you know you will use them.
Is there any free backtesting software?
Yes, and several free options are genuinely good. QuantConnect offers unlimited backtests on its free tier if you code. TradingView runs its Strategy Tester on the free plan. Portfolio Visualizer analyzes up to 15 assets free. Fidelity Strategy Testing and TradeStation strategy backtesting are free with a funded account. The usual limits are saving, exporting, and data depth.
How much does backtesting software cost?
Free to about $400 per month, and most people belong at the low end. Free tiers exist at QuantConnect, TradingView, Portfolio Visualizer, Fidelity, and TradeStation, though QuantConnect's free node is capped at 200 backtests a day and cannot trade live at all, as the <a href="/quantconnect-pricing">QuantConnect pricing</a> breakdown shows. Plain-English benches start at $19 per month, portfolio tools run $30 to $55, chart-condition platforms $59 to $349, and AmiBroker is a one-time $299 to $499 plus data. Watch for AI features billed on top of the plan.
Does backtesting work?
Backtesting reliably tells you what a rule did in the past and reliably overstates what it will do next. Measured edge decays sharply out of sample: McLean and Pontiff, writing in the Journal of Finance in 2016, found that returns to 97 published stock-return predictors were 26% lower out of sample and 58% lower after publication.
Is backtesting a waste of time?
It becomes a waste of time when you use it to search for a winning configuration rather than to test an idea you already hold. Bailey, Borwein, Lopez de Prado and Zhu showed in 2014 that with only five years of daily data, trying more than about 45 independent configurations will almost certainly produce an in-sample Sharpe of 1.0 whose expected out-of-sample Sharpe is zero.
Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.
Your next idea deserves a verdict, not a hunch.
Bring a thesis or a ticker. AgentTrading restates the rule, shows the evidence, runs 20+ years of history, and stamps an honest verdict. You decide.
Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.