Best Backtesting Software for ETF Rotation Strategies

The best backtesting software for an ETF rotation strategy is the one that lets you describe the rule in plain terms, tests it on at least 20 years of dividend-adjusted prices, charges trading costs on every switch, and compares the result with simply holding the index. A no-code builder such as Composer suits you if you want to automate the rotation afterwards. A plain-English tester such as AgentTrading suits you if the question is whether the rotation has ever paid. Spreadsheet-based and code-based tools work, but cost you weekends.

This guide is for someone with a specific idea in hand, such as "hold whichever of three ETFs had the strongest last six months, move to bonds when the market is below its 200-day average", who wants to pay for the right tool once. Prices below were read from vendor pricing pages, and the Composer figures on November 6, 2026.

What an ETF rotation backtest has to get right

A rotation strategy trades often by design, so it exposes every weakness in a backtester. Five things decide whether the answer you get is usable.

• Adjusted prices. ETFs pay distributions, and bond and dividend ETFs pay a lot of them. A test on raw prices understates what you would have earned and distorts any rule that compares ETFs against each other.
• Trading costs on every switch. A monthly rotation can trade a dozen times a year. A test with zero costs flatters it. The question is what is left after costs, not before.
• History that spans bad markets. A rotation tested from 2010 has barely met a bear market. You want 2000 to 2002 and 2008 inside the window, which means 20 years or more.
• A fair benchmark. The rotation has to beat buying the index and doing nothing, including in drawdown terms. Many rotations win on return only because they took more risk.
• A trail you can audit. You should be able to see which ETF was held in which month and why. A single curve you cannot inspect is not evidence.

One limit applies to every tool. An ETF only has history from its launch date, so a rotation among newer funds cannot be tested through 2008. Pick a rotation set where the oldest and youngest funds both give you the window you care about, or substitute a longer-lived fund and say so in your notes.

The main ways to backtest a rotation

When Composer is the right choice

Composer is built around exactly this kind of strategy. You assemble conditions and asset choices in a visual editor, test them, and run the strategy automatically through the platform. The Starter plan is $0 for one automated strategy of up to 200 nodes, with unlimited backtesting in the editor, so you can build the rotation and test it before you pay anything. A node is one block in the decision tree, and a rotation across several ETFs with a trend filter can use a fair number of them.

The honest cost is not the subscription. Live trading means opening a brokerage account with Composer, FINRA and SEC regulatory fees apply to sales, and Pro, which adds API access and a much higher node ceiling, is billed yearly. If you want the strategy to run itself, those are fair trades. If you only want to know whether the idea works, you are signing up for more than you need. Our Composer review goes through the terms line by line.

When a plain-English tester is the right choice

AgentTrading starts from the sentence you would say out loud. You type the rule, the bench restates it as an explicit rule card, and you confirm it before anything runs. The test uses 20+ years of split- and dividend-adjusted daily data, charges a default 0.1% cost per trade that you can adjust on Pro, and returns a plain verdict against buy and hold. That verdict includes UNDERPERFORMED when the simple index wins, which for a surprising number of rotations is the true answer.

This is the tool for the stage before automation. It places no orders and holds no money, so there is nothing to connect and nothing to unwind. Starter is $49 a month, or $288 a year. Fixed-weight baskets with scheduled rebalancing are covered on portfolio backtesting, trend and moving-average rules on moving average crossover backtesting, and momentum rules on momentum backtesting. Always read the restated rule card, because a rotation among several ETFs has more moving parts than a single-ticker rule and you want to be certain the tool understood yours.

A sensible way to test a rotation before paying anyone for anything

1. Write the rule as one sentence. Universe, ranking measure, how many you hold, how often you rotate, and the fallback asset. If you cannot fit it in a sentence, it has too many parameters to trust.
2. Test it over a window that includes at least two bad markets. The window length matters more than the exact dates.
3. Charge costs. Run it once at the default cost and once at double. A rotation that only works at zero cost does not work.
4. Look at drawdown, not just return. If the maximum drawdown is close to the index, the rotation added trading but not protection.
5. Change one parameter and rerun. If moving the lookback from six months to seven flips the result, you have found noise. Our guide to backtesting a momentum strategy shows the check.
6. Only then decide whether to automate it. Automation is the expensive step. A rule that survived the first five steps is worth paying to run.

A note on the data under the test

Every result above is only as good as the price history beneath it. A missing distribution, a bad split or a stale series changes which ETF looks strongest in a given month, and rotation rules are very sensitive to that ranking. If you are assembling your own data, a data observability service that checks prices for gaps and anomalies catches the silent errors before they reach a backtest.

What does ETF rotation backtesting software cost?

You can test a rotation for $0 on Composer's Starter plan, or for $49 a month on AgentTrading Starter. Automating it costs more, from $10 a month on Composer Advanced to $32 a month billed yearly on Pro, plus regulatory fees on sales. For the wider price picture across the category, see how much stock analysis software costs.

Which should you pick?

AgentTrading is research and strategy analysis only. It does not execute trades, connect to a brokerage or give personal advice, and a backtest is hypothetical and benefits from hindsight. Get started to run your rotation rule on 20+ years of adjusted data and see the verdict.