The Best Free Backtesting Apps in 2026: Replay Tools and Code Testers Compared Honestly

The best free backtesting app is TradingView if you test setups by eye on a chart, MetaTrader 5 if you trade forex and can write your rules as an Expert Advisor, and QuantConnect or a Python library like Backtesting.py if you want to test across years of data in code. FX Replay's free plan is fine for drilling one setup in a browser, but its caps are tight. None of them can tell you a strategy will make money; what a free backtest can do is show you that a rule loses money before you pay the market to find out.

This comparison covers what each free plan actually allows, where each one quietly limits you, and a worked example of the arithmetic a backtest has to survive before it means anything.

Quick comparison

What "free" means for a backtesting app

A backtest needs three things: historical prices, a way to apply your rules to them without seeing the future, and a record of every trade so you can do the math afterwards. Free plans ration one of those three. TradingView rations history and indicators, FX Replay rations sessions and trades, and QuantConnect rations compute and log size. The Python libraries ration nothing, but you have to bring the data and write every rule yourself.

So the useful question is which limit you can live with. A swing trader testing a daily-chart pullback rule barely notices TradingView's free caps. A scalper testing a five-minute opening setup runs into them on the first afternoon. A forex trader who already has an MT5 broker account gets a full coded tester at no cost, but only if they can write MQL5 or are willing to learn it.

There is also a split between the two styles of testing. Manual backtesting means replaying a chart bar by bar, making each decision as if it were live, and logging the trade. Coded backtesting means writing the rules as a program and letting the computer apply them across the whole history. Manual testing teaches you to see the setup, which is the skill you need when you trade. Coded testing gives you hundreds of trades in seconds, but only for rules precise enough to program. We covered both methods step by step in how to backtest a trading strategy; this post is about which free tool fits which method.

TradingView: best free backtesting app for chart traders

TradingView is where most retail traders already read charts, which is the main reason to backtest there: the drawing tools, indicators and layout you test with are the ones you will trade with. The free Basic plan includes two backtesting tools.

The first is Bar Replay. You pick a past bar, the chart hides everything to the right of it, and you step forward one bar at a time or let it play at a speed you set. You can draw levels and add indicators as the replay runs, and TradingView's Bar Replay help page lists how deep the history goes: daily AAPL data from December 1980, one-minute AAPL from January 2000, one-minute EURUSD from November 2001, and one-second bars from August 2022 for the symbols it lists.

How far back you can replay on lower timeframes depends on the plan. The TradingView pricing page lists replay on daily and higher timeframes with full history on every plan, while minute-level replay history is capped on the lower tiers (180 days, then 365 days) and unlimited higher up. Check the current table for your tier before you plan a five-minute test, because a replay that cannot reach the date you picked shows "Data point unavailable" and stays on the higher timeframe.

The help page also lists limits that matter for testing. Renko, Kagi, Point and Figure, Range, Line Break and tick-based charts do not support Bar Replay. You cannot create new server-side alerts during a replay. And trading orders, including Paper Trading orders, execute on real-time data, so you cannot place paper trades against the replayed candles. You log replay trades yourself, in a spreadsheet or journal.

The second tool is the strategy tester, which runs a strategy written in Pine Script and reports net profit, drawdown, win rate and the trade list. On the free plan it runs on the bars loaded in the chart, and the pricing page lists 5,000 historical bars for Basic. Five thousand daily bars is about 20 years, which is plenty. Five thousand five-minute bars is roughly 65 regular US stock sessions, about three months, which is thin for an intraday rule. Deep Backtesting, which tests across the full history up to two million bars and one million trades, is available on Premium and higher only.

The free plan also allows two indicators per chart. A pullback rule using a 20 EMA and RSI fits; a rule that also needs volume and ATR does not, unless you combine them into one Pine Script indicator.

Use it for discretionary setups on daily or four-hour charts. On the free plan, expect to log replay trades by hand and to hit the caps on intraday history and indicator count.

MetaTrader 5 Strategy Tester: best free tester for forex rules you can code

MetaTrader 5 is free to download for Windows, macOS and Linux, with mobile apps and a browser terminal, according to MetaQuotes' download page. It is the platform many forex brokers hand you when you open an account.

Its Strategy Tester runs an Expert Advisor (a trading program written in MQL5) against historical quotes. You choose how finely it simulates price inside each bar: "every tick," which MetaQuotes calls the most accurate mode, "1 minute OHLC" as a faster middle ground, or "open prices only" for a quick rough pass. The tester also optimizes inputs, running the same EA with many parameter sets and using a genetic algorithm to cut the number of combinations, and it can split the data into an optimization segment and a forward segment to check whether the best settings hold up on data they were not tuned on. Robots can trade several symbols in one test, so a rule that depends on the correlation between two currency pairs can be tested properly.

Visual testing mode draws the EA's trades on the chart as the test plays, and you can slow it down or pause it. That makes it useful for checking that the code does what you meant. A coded rule that enters on the wrong bar is the most common bug in a first EA, and you only see it by watching.

The catch is the code. If your rule is "buy the first pullback to the 20 EMA after a higher high," you have to define the higher high, the pullback and the touch in MQL5, and every vague word in your rule becomes a decision you must make. That discipline is useful in itself. The other catch is data: the tester uses quotes from your broker's server, so a test on one broker can differ from the same test on another, especially around news and the daily rollover when spreads widen.

MT5 suits forex traders with mechanical rules and the patience to code them. It offers nothing for a setup you can only judge by eye, and results vary with the broker's data.

FX Replay: best free browser replay for drilling one setup

FX Replay is a browser tool built for manual backtesting: pick a market and date, replay candle by candle, and place simulated trades on the replayed chart, which TradingView's replay does not allow. Its pricing page lists the free Beginner plan as "free for everyone," with no trial period and no automatic upgrade.

The free limits are tight: two backtesting sessions, one indicator, 50 trades per session and one week of data retention. Fifty trades is enough to learn whether you can spot a setup consistently, and too few to judge whether it makes money (the sample-size section below shows why). Paid plans add sessions, indicators, data retention and seconds-level data; the page quotes the Intermediate plan at $15 a month on the plan card and $17.99 billed monthly in the FAQ, so confirm the current price before you subscribe.

We compared FX Replay with broker demos and other replay tools in more depth in our guide to free forex simulators online.

It is a good free place to practice one setup with trade entry on the replay, though the caps end a serious test early.

QuantConnect: best free cloud backtester with data included

QuantConnect is a cloud platform for coded strategies. Its pricing page lists the free plan as including unlimited backtesting and equity, index, forex, crypto, futures and options data, with a web IDE that has autocomplete and a debugger. That data is the main reason to pick it: finding clean, split-adjusted stock history for free is harder than writing the strategy.

The free limits are about compute and output rather than data: one backtest node, one research node, no live trading node, a 10KB backtest log, 10,000 orders per backtest and 4,000 plot points. One node means one backtest runs at a time, so a parameter sweep is slow. The 10KB log fills quickly if you print every trade, so you rely on the built-in report.

The learning curve is real. You are writing a program against an API with its own conventions for data, orders and scheduling, and a first strategy usually takes a few evenings before the first clean backtest.

Pick QuantConnect if you code and want multi-asset data without hunting for it. Expect a steep start, slow parameter sweeps on one node, and no live trading unless you pay.

Backtrader and Backtesting.py: best free option if you write Python

If you already know some Python, two open-source libraries cover most retail backtesting needs on your own machine with no plan limits.

Backtrader describes itself on its site as "a feature-rich Python framework for backtesting and trading." Its documentation covers multiple timeframes in one strategy, resampling minute data into larger bars, data replay, futures rollover, TA-Lib integration and live trading connections for Interactive Brokers and Oanda. It is the more complete of the two and the more verbose to write.

Backtesting.py is lighter. Its documentation describes it as built on Pandas, NumPy and Bokeh, working with any instrument that has candlestick data, compatible with TA-Lib and pandas-ta, and shipping an optimizer that tests many variants and draws heatmaps of the results. A simple moving-average crossover runs in about twenty lines, and the output chart shows every trade on the price series. The page is plain about two things: "You need to know some Python to effectively use this software," and past performance is not indicative of future results.

With either library, the data is your job. Free daily stock data is easy to get; free intraday data is limited and often unadjusted for splits and dividends, which can produce fake gaps that look like trades. Check a few known split dates in your data before trusting any result.

Choose a library if you code and want no caps and full control. No data comes with either, and every rule, cost and fill assumption is yours to get right.

Forex Tester and other paid desktop tools

Forex Tester is the best-known paid desktop program for manual replay, with tick data and simulated order entry. Its pricing page blocked our fetch for this article, so we are not quoting a price; check forextester.com directly. Paid tools earn their cost if you replay hundreds of hours a year and want tick data and realistic spread modeling in one place. For a first test of a setup, the free options above are enough to find out whether the idea deserves that spend.

A worked example: testing one setup on a free plan

Say you swing trade US stocks and want to test this rule: in a daily uptrend (price above a rising 50-day moving average), buy when price pulls back to the 20-day EMA and the next candle closes above the prior candle's high. Stop under the pullback low. Target two times the risk, so every win is +2R and every loss is -1R.

On TradingView Basic, this fits the free plan. Two indicators (the 50 SMA and 20 EMA), daily bars, full history. You replay ten liquid stocks over several years and log every qualifying setup by hand: date, entry, stop, target, result. After a few weekends you have 60 trades. Twenty-four won and thirty-six lost, a 40% win rate.

The expectancy, the average result per trade in R, is win rate times average win minus loss rate times average loss:

0.40 x 2R minus 0.60 x 1R = 0.80R minus 0.60R = +0.20R per trade.

That looks positive, but the replay filled every entry at the exact candle close and every stop at the exact stop price. Live, you pay commissions where they apply, the bid-ask spread, and slippage when a stop triggers in a fast move or the stock gaps through it overnight. If those costs average 0.10R per trade, a placeholder you should replace with your own numbers, the edge drops to +0.10R. If they average 0.20R, it is gone.

Bar chart of expectancy per trade for a strategy with a 2R target and 1R stop, before and after 0.10R of costs per trade. At a 30% win rate expectancy is -0.10R before costs and -0.20R after. At 35% it is 0.05R and -0.05R. At 40% it is 0.20R and 0.10R. At 45% it is 0.35R and 0.25R. At 50% it is 0.50R and 0.40R.

The same test on FX Replay's free plan would stop at 50 trades in a session, and the one-indicator cap means dropping either the 50 SMA or the 20 EMA, which changes the rule. In Backtesting.py the rule takes an afternoon to code, runs across 20 years of daily data for a few hundred stocks, and returns hundreds of trades. The coded version will also catch setups you would have skipped by eye, and miss ones your eye would have taken, because "pulls back to the EMA" needs an exact definition (within 0.5% of it? a wick touch? a close below?). Running both, the manual version first, tells you whether the edge lives in the rule or in your judgment.

How many trades a free backtest needs

Sixty trades at a 40% win rate sounds like evidence. It is weak evidence. The standard error of a win rate is the square root of p times (1 minus p) divided by the number of trades. At 40% and 60 trades, that is about 6.3 percentage points, so a rough 95% range for the true win rate runs from about 27.6% to 52.4%. The expectancy chart above shows that range covers both a losing strategy and a very good one.

Bar chart of the approximate 95% margin of error on a measured 40% win rate by number of trades. 20 trades: plus or minus 21.5 percentage points. 50 trades: 13.6. 100 trades: 9.6. 200 trades: 6.8. 500 trades: 4.3.

This is where free plans differ most in practice. A coded tester gets you to 200 or 500 trades easily if the rule fires often enough. A manual replay gets you there slowly, and FX Replay's free 50-trade sessions get you there only across several sessions, and the plan allows two. For a manual test, 100 trades is a reasonable minimum before you trust the direction of the result, and even then treat the exact win rate as a rough figure.

The trades also need to span different conditions. One hundred trades from a single strong bull year tells you how the rule behaves in a bull year. Spread the replay across a trending year, a choppy year and a falling year, and log which one each trade came from.

Five ways a free backtest flatters you

Perfect fills. Every replay and most testers assume you got the price you wanted. Add a per-trade cost for spread, commission and slippage, and test what happens when a stop fills a few ticks worse or the stock gaps past it.

Hindsight in manual replay. If you have seen the chart before, you remember where it went. Start replays on dates you have not studied, hide the symbol name if your tool allows it, and write the rule down before you start so you cannot adjust it mid-test.

Over-tuned settings. An optimizer will always find a moving-average length that worked best on the past. A 17-period EMA beating the 20 by a small margin is noise. MT5's forward-testing split and a simple rule of keeping settings round and few are the free defenses: test on one stretch of data, then check the same settings on a later stretch you did not touch.

Survivorship in the symbol list. Testing today's best-known stocks over the last ten years tests companies that survived. A rule that buys pullbacks in strong stocks looks better when the stocks that collapsed are not in the list.

Data errors. Unadjusted splits, missing bars and bad ticks create trades that never existed. Spot-check the biggest winners and losers in any coded backtest against a chart. If the best trade in the test is a split date, the test is broken.

Every one of these comes from the same place: a backtest is prepared with hindsight and cannot fully account for live execution. A backtest that survives costs, enough trades and an untouched data segment earns a forward test on a demo or very small account; it has not earned real size. Our guide on how long to paper trade before going live covers that next step.

Where Quant AI fits

Quant AI is not a backtesting app. It reads a screenshot of a stock, crypto or forex chart and marks the trend, support and resistance levels, and patterns it finds. In a manual backtest that makes it a second opinion on the bar in front of you: pause the replay, take a screenshot, and check whether the app reads the trend and the level the same way your rule does. When it disagrees, write down why; those disagreements are often where a vague rule needs a sharper definition.

What it does not do: run your rules across history, log trades, compute expectancy or tell you a strategy works. That part stays with the tester and with you.

How to choose

• You read charts and trade daily or four-hour setups: start with TradingView Basic's Bar Replay and a spreadsheet log. Upgrade only if the indicator or intraday caps block the rule you are testing.
• You trade forex with mechanical rules: use the MetaTrader 5 Strategy Tester through your broker, and use visual mode to check the code enters where you think it does.
• You want to drill one setup in a browser with trade entry on the replay: FX Replay's free plan, knowing you will outgrow 50 trades per session.
• You code and want data included: QuantConnect's free plan.
• You code and want no limits: Backtesting.py for speed of writing, Backtrader for multi-timeframe and live-trading connections.
• You trade intraday and replay many hours a week: test the idea free first, then consider a paid replay tool or a paid TradingView tier with deeper minute history.

Frequently asked questions

What is the best free app for backtesting trading strategies?

For most chart traders it is TradingView, because Bar Replay and the strategy tester are on the free Basic plan and run on the charts you already use. For coded forex rules it is the MetaTrader 5 Strategy Tester. For coders who want multi-asset data included, QuantConnect's free tier.

Are there any genuinely free backtesting tools?

Yes. MetaTrader 5, Backtrader and Backtesting.py have no paid tier for the testing itself. TradingView, FX Replay and QuantConnect have free plans that work for real tests within their caps: loaded bars and two indicators on TradingView, two sessions of 50 trades on FX Replay, one backtest node on QuantConnect.

What is the best free backtesting app for Android?

Serious backtesting is still easiest on a desktop or in a browser, because replaying hundreds of bars and logging trades on a phone screen is slow. On Android, the practical route is a browser-based tool such as FX Replay, or TradingView, opened on a tablet if you can. Use the phone to review your log, not to run the test.

What is the best free backtesting software for forex?

The MetaTrader 5 Strategy Tester for coded rules, since most forex brokers already supply MT5 and it tests on their quote history. For manual forex replay, TradingView's Bar Replay has one-minute EURUSD history back to 2001, with the minute-level depth you can reach set by your plan.

Can ChatGPT backtest a trading strategy?

It can write the code for a backtest, for example a Backtesting.py script, but it does not have reliable price history of its own, so the numbers have to come from data you supply and code you run. Read the code it writes: check how it fills entries, whether it includes costs, and whether any line uses a future bar to make a past decision. Those are the bugs that make an AI-written backtest look better than the strategy is.

How many trades does a backtest need?

At least 100 for a manual test before you trust the direction of the result, and 200 or more before you trust the win rate within about seven points. Spread them across trending, choppy and falling markets.

Is a profitable backtest enough to trade the strategy live?

No. It earns a forward test: the same rules on a demo or a very small live account, logged the same way, for long enough to collect a meaningful number of trades. Backtests are prepared with hindsight and fill more kindly than a live market, so expect live results to be worse, and size the first live trades as if the edge might not exist.

The bottom line

Pick the free tool that matches how you test. TradingView for chart setups, MetaTrader 5 for coded forex rules, FX Replay for short browser drills, QuantConnect or a Python library for coded tests across years of data. Whichever you use, the result only counts after costs, after enough trades, and after the settings survive data they were not tuned on. Trading carries real risk of loss, and a good backtest lowers the odds of trading a broken rule; it does not remove the risk.