Backtesting software ranges from a tab on a chart to a code library you run yourself. Here is how the main tools differ, read from their own pages, which one fits whom, and the traps that follow you into every one of them.
November 16, 2026
12
min to read
The short answer
The best backtesting software depends on how you work. TradingView suits chart-first traders who can write a little Pine Script, QuantConnect and Python libraries suit coders, MetaTrader 5 and NinjaTrader suit forex and futures traders, and TrendSpider and Ethy let you describe a strategy in plain words. Several have a free option.
A backtest replays a trading rule on past prices and shows what it would have done. Every tool below does that. What differs is where the rule lives and who has to write it, and that splits the market into four kinds.
None of the four is better in general. A coder who wants a thousand variants of one idea and a trader who wants to check a hunch before lunch need different things. If you are new to the whole idea, start with how to backtest a trading strategy, which covers the method these tools all rest on.
Here are the nine side by side. Every row comes from the tool’s own pages, read on 16 November 2026. Where two of a tool’s pages disagree, the note under the table says so.
| Tool | Who it suits | Markets | Coding needed | Free option | What it does well |
|---|---|---|---|---|---|
| TradingView Strategy Tester | Chart-first traders who already use TradingView | Whatever symbols you can chart | Only for your own rule (Pine Script); built-in and community strategies need none | Yes (a free Basic plan) | Results sit under the chart, with a Metrics tab and a list of trades; forward testing as new data arrives |
| QuantConnect (LEAN) | Coders who want a full research and live-trading stack | Equities, indexes, forex, crypto, futures and options data on the free plan | Yes, Python or C# | Yes (free plan, no live trading) | Open-source engine built for research, backtesting and live trading; multi-asset portfolio modeling |
| backtesting.py | Python users who want a small, quick library | Any instrument with candlestick data: forex, crypto, stocks, futures | Yes, Python | Yes (open source, AGPL-3.0 license) | Lightweight, built-in optimizer with heatmaps, interactive charts |
| Backtrader | Python users who want one framework from backtest to broker | Whatever data you load: CSV, pandas, Yahoo and others | Yes, Python | Yes (open source) | Brokers, commission schemes, analyzers and sizers built in; live links to Interactive Brokers, Oanda and Visual Chart |
| vectorbt | Python users who test many variants of an idea fast | Whatever data you load into pandas | Yes, Python | Yes (free to use; a paid PRO membership also exists) | Vectorized: a very large grid of variants tested at once |
| MetaTrader 5 Strategy Tester | Forex, stock and futures traders who run Expert Advisors | Forex, stocks and futures, with history from your trade server | Yes, to write an Expert Advisor; or buy or commission one | Yes (a free download) | Five tick generation modes, tests across several instruments, optimization on local and remote agents, forward testing |
| TrendSpider Strategy Tester | Technical traders who want no code | Stocks, ETFs, futures, forex and crypto; not options or indices (see the note below) | No: a visual editor or plain English | No (a paid trial only) | Sharpe, Sortino and max drawdown in the report; a group test across up to 500 symbols at once |
| NinjaTrader | Futures traders | Futures are what its homepage features | Not with the Strategy Builder, which writes the NinjaScript for you; C# for your own code | Yes, free for backtesting and simulation | Market Replay, plus walk-forward optimization |
| Ethy | Crypto and Hyperliquid traders who want no code | Spot tokens, and every perp on Hyperliquid’s main exchange and on trade.xyz (stocks, commodities, indices) | No: describe it in plain English | Yes (Free plan: the shortest window per timeframe) | Runs the exact rules that would deploy; train and held-out test windows; one tap to paper or live |
One caveat on TrendSpider: its help center and its product page say options and index data can’t be backtested, while its learning center says the same strategy logic runs across stocks, ETFs, options, futures, forex and crypto. The pages disagree on options, so check your market before you pay.
Pine Script strategies “simulate trades across historical and realtime bars,” according to the Pine Script docs, and the results appear in a report under the chart with a Metrics tab and a trade list. By default the order-fill emulator uses only chart data, so it has to assume the path price took inside each bar. Premium and Ultimate plans can switch on lower-timeframe detail to sharpen that. The default range keeps the individual data of the latest 9,000 trades, and Deep Backtesting keeps all of them. The report shows one strategy at a time. The support article adds two things: you can load a built-in or community strategy instead of writing one, and the report keeps updating as new data arrives, which TradingView calls forward testing. If you want Claude or ChatGPT to write the script, see Pine Script with AI.
LEAN is described in its docs as a “radically open-source algorithmic trading engine,” built for strategy research, backtesting and live trading, and “multi-asset with full portfolio modeling.” Examples are in Python, with C# also available. The free plan offers unlimited backtesting on one backtest node, with minute, hour and daily data. It has no live trading, and tick and second data are not on it. Live trading comes with the paid plans. Everything here is written in Python or C#, so it suits coders best.
backtesting.py calls itself lightweight and fast, works on any instrument you have candlestick data for, and includes an optimizer with heatmaps plus stop-loss and take-profit settings. Its GitHub repository is open source under the AGPL-3.0 license. Backtrader is a broader framework, with brokers, commission schemes, analyzers, sizers and live-trading links to Interactive Brokers, Oanda and Visual Chart. vectorbt takes a different route: it packs many strategy variants into one array so it can test “many thousands of strategies in seconds,” in its own words. Its repository says anyone may use it for free, under an Apache 2.0 license with the Commons Clause, and a paid PRO membership adds more features. You bring or fetch the price data yourself for all three, and you need some Python.
The tester runs Expert Advisors, the automated programs of MetaTrader. It offers five tick generation modes, from “every tick” (“the most accurate but the slowest”) to “open prices only,” can model network delay, tests strategies that trade several instruments at once, and runs optimization across local and remote agents. It also supports forward testing on the later part of the period. MetaTrader 5 describes itself as a multi-asset platform for forex, stocks and futures, its robots can be built, bought or ordered from programmers, and the platform is a free download.
Its help center says “any trader who can describe a strategy in words to a friend” can use the Strategy Tester. It is also candid about limits: fees are assumed to be zero until you set them, execution is assumed to be perfect with no slippage, a test runs on one timeframe, and you can’t go long and short in the same strategy. Its product page lists Sharpe and Sortino ratios and maximum drawdown in the results, and a Group Strategy Tester that compares strategies across up to 500 symbols at once. Its learning center adds that how far back one test can reach depends on your plan.
NinjaTrader’s homepage says it is “always free to use for advanced charting, backtesting and trade simulation,” with Market Replay to “review and test strategies using historical market data.” Its NinjaScript docs describe strategies as “automated, backtestable trading systems” with “optimization and walk-forward analysis.” In a walk-forward run you set an in-sample period for tuning and an out-of-sample period that tests the result. NinjaScript is C#, but the same docs say the Strategy Builder generates it for you if you would rather not write code by hand. Futures are what its homepage features.
You describe the idea in plain language, preview the rules, and backtest the exact spec that would deploy. It covers spot tokens and every perp on Hyperliquid’s main exchange and on trade.xyz, with windows up to 365 days depending on the timeframe. Results include the equity curve, drawdown, Sharpe, Sortino, a comparison with buy and hold, and every entry and exit marked on the chart. The optimizer ranks candidates on a training window and checks each one on a held-out test window. The limits are real too: it tests Ethy’s own markets only, so you can’t point it at a forex or futures broker’s data, and the free plan runs only the shortest window for each timeframe (180 days on the daily). More in Ethy’s backtesting page, or see how a ChatGPT or Claude backtest compares.
It depends on what you are willing to type. If you code, the Python libraries and QuantConnect’s free plan cost nothing and have no ceiling on how many tests you run. If you trade futures, NinjaTrader says backtesting is free, and its Strategy Builder writes the code for you. If you trade forex through a broker, MetaTrader 5 is a free download. If you want a free start without code, TradingView’s Basic plan runs built-in and community strategies (your own rule needs Pine Script), and Ethy’s Free plan takes plain English. TrendSpider is the one tool here without a free plan, only a paid trial.
Free often means a shorter history, fewer runs or no live link. That is usually fine while you are learning. A free tool that tests a short window is still better than no test, as long as you remember how little a short window proves.
Work through four questions, in this order.
Run the same simple rule through two tools. If the trade lists disagree, find out why before trusting either. The cause is usually a fill assumption or the data, and learning that is worth more than any feature on a comparison page.
The software only runs the test you give it. These three traps live in the test, so they follow you everywhere. The full method is here; this is the short version for tool shoppers.
A rule that uses information it would not have had at the time, such as a bar’s close used to buy at that bar’s open, looks brilliant and fails live. Each tool handles fills differently: TradingView’s default emulator fills on historical bars after a bar closes, and MetaTrader’s “open prices only” mode uses only each bar’s open price. Read how your tool fills orders, and have the rule act on the next bar, not the current one.
Optimizers are the best feature of these tools and the most dangerous. Tune enough settings on one stretch of history and some combination will fit it perfectly by chance. vectorbt’s pitch of thousands of variants in seconds, MetaTrader’s optimization and backtesting.py’s optimizer all make this easy. The defense is to test on data the tuning never saw: MetaTrader’s forward testing, NinjaTrader’s walk forward optimization, TrendSpider’s test across many symbols, and Ethy’s held-out test window are each a version of it.
A backtest with no trading costs flatters every strategy that trades often. TrendSpider’s own help page says fees are assumed to be zero until you change them. Whatever tool you use, find the setting for costs and slippage, set it, and assume a live fill will be worse than a simulated one.
For coders, the Python libraries and QuantConnect’s free plan. For futures, NinjaTrader, which says backtesting is free. For forex, MetaTrader 5, a free download. For no code, TradingView’s Basic plan (built-in and community strategies; your own needs Pine Script), NinjaTrader’s Strategy Builder, or Ethy’s Free plan (plain English, shortest window per timeframe). TrendSpider has no free plan.
Not any more. TrendSpider and Ethy build the rule from a visual editor or plain English, and NinjaTrader’s Strategy Builder writes the NinjaScript for you. TradingView runs built-in and community strategies, but a rule of your own needs Pine Script. MetaTrader 5 needs an Expert Advisor you write, buy or order. QuantConnect needs Python or C#, and the Python libraries need Python.
No. It replays a rule on past prices. A good result means the rule fit that history, not that it will work next week. Test on data you did not tune on, and treat every result as an estimate.
Ethy covers spot tokens and every perp on Hyperliquid’s main exchange with no code. QuantConnect’s free plan lists crypto data, and the Python libraries take any candles you load. TradingView tests whatever symbol you can chart. Check the history length for your coin.
Checked 16 November 2026 on each vendor’s page; prices move. TradingView: Basic is free; Essential, Plus, Premium and Ultimate list at $14.95, $34.95, $69.95 and $239.95 a month, less billed yearly. TrendSpider: no free plan; promotional first-year prices of $32.94 to $74.90 a month. QuantConnect: a free plan, with paid plans on its pricing page. NinjaTrader: free for backtesting. MetaTrader 5: a free download. Ethy: Free $0, Core $15 a month. Python libraries: free.
Trading involves risk, including the loss of your money. Ethy AI is a research and trading tool and does not give investment advice or recommendations. Backtests are simulations on historical data and do not predict future results.
Deploy your AI trading agent in one click.
