Windows portfolio backtester scripted in AFL, sold as a perpetual licence.
Backtesting software and platforms
Every card that can replay a strategy or an allocation against history, sorted by what it is and by what it assumes about fills and data.
Last updated
One flag, five different purchases
A backtesting flag on a card says the product can replay something against history. That
covers at least five things nobody would put side by side:
- A tester inside a charting platform — TrendSpider, the Pine Script strategy tester in TradingView, the OddsMaker in Trade Ideas, ProBacktest in ProRealTime.
- A desktop platform with its own language — AmiBroker in AFL, MultiCharts in PowerLanguage, NinjaTrader in NinjaScript, WealthLab in C#.
- A no-code builder — Composer, Tradetron, Option Alpha, and the strategy generators Build Alpha and StrategyQuant X.
- An engine you call from code — Backtrader, LEAN, vectorbt, NautilusTrader.
- An allocation backtest or a withdrawal replay — Portfolio Visualizer, testfolio, Backtest by Curvo, cFIREsim, FI Calc.
Whoever searches for "backtesting software" usually wants the first or the third: a product with a screen. The libraries make up the backtesting frameworks section, along with AmiBroker and WealthLab, where the backtest is the thing being bought. The rest are filed by their main job, under no-code strategy builders, charting and screeners, trading platforms and retirement planning.
What decides whether a result means anything
The fill. Composer tests on daily adjusted closes with a default slippage of one basis point and no intrabar fills. Backtesting.py fills a market order at the next bar's open unless told to use the close. LEAN's default brokerage model charges the Interactive Brokers commission schedule and assumes zero slippage. Backtrader can slip by a percentage or a fixed amount and cap a fill at a share of the bar's volume; Option Alpha fills at the mid and does not model assignment; Build Alpha asks for slippage and commission per strategy rather than assuming them.
The data, and whose it is. TrendSpider includes real-time US equity data for non-professionals and reaches 50-plus years where a symbol has it, with each test capped by candle count. Build Alpha bundles history back to 2006, intraday from one to 240 minutes. AmiBroker, MultiCharts, vectorbt and NautilusTrader ship none, so the feed is a second bill. LEAN ships samples, and its US Equity Security Master of splits, dividends and ticker changes costs $600 a year on a Researcher seat.
Survivors and corporate actions. A universe stripped of its delisted names tests only the companies that made it — survivorship bias. Portfolio123 keeps delisted stocks, WealthLab's bundled DataSets track index membership as it was on the day, QuantConnect's AlgoSeek equities from 1998 are survivorship-bias-free, and QuantRocket and ORATS retain delisted tickers. CSI Data sells delisted stocks only as a premium add-on. ETFreplay runs on total return with distributions reinvested; Curvo assumes reinvestment whether or not the fund distributes. The sourcing is in how to backtest on a survivorship-free universe.
The language. Clicks in Composer and TrendSpider; Pine Script, AFL, PowerLanguage, ProBuilder or NinjaScript on the platforms; Python or C# in the engines. A proprietary language keeps the strategy where it was written: ProBuilder code runs nowhere but ProRealTime, and AFL nowhere but AmiBroker. Build Alpha, StrategyQuant X and Adaptrade Builder cross that line by emitting code for other platforms, and all three cards say that overfitting is built into searching for rules.
Where the free plan stops
TradingView's free plan loads 5K bars; paid plans raise that to 40K, and deep backtesting opens at Premium. TrendSpider has no free tier: its $82 plan tests nothing finer than two-hour bars over 2,000 candles, and one-minute testing starts at $183. Trade Ideas sells the scanner at $127 and keeps the OddsMaker for the $254 tier. ProRealTime's free web version runs no backtests.
Composer works the other way: backtests are free, and the $40 Trading Pass buys automated execution and unlimited backtesting. Tradetron charges ₹20 a run. Portfolio123 gives a free account 30 days of backtests, then prices its plans by years of history, from 5 to 20. QuantConnect's free tier has one backtest node, no second or tick data and a 10,000-order ceiling per backtest. NinjaTrader's backtesting needs no funded account; the free MultiCharts is a simulator on sample data.
For allocations, Portfolio Visualizer limits free ticker backtests to ten years, testfolio's free tier keeps full history and PortfoliosLab stops at five years. cFIREsim, FI Calc and Curvo have no paid tier at all.
The cards here for a different reason
- Retirement planners — FIREproof, ProjectionLab, Pralana Gold, Pralana Online, WealthTrace. They replay a whole plan, taxes included, through past sequences or simulated ones; as the FIREproof card notes, overlapping windows are not independent trials, so a success rate is not a probability.
- Trade journals — TraderSync, TradesViz, TradeZella, each with a market replay or a backtester beside the journal, for practising on history rather than recording it.
- Research sites with a backtest tab — Uncle Stock, GuruFocus, YCharts, ORTEX, Quiver Quantitative, WhaleWisdom, Snowball Analytics, The Dividend Tracker: a test run over the site's own data, such as a saved screen or a 13F fund held as a portfolio.
- Options data with a tester attached — ORATS sells an intraday backtester as an add-on; OptionNET Explorer replays positions by hand.
- Data for somebody else's engine — CSI Data exports into whatever backtester you run, SimFin ships as-reported statements for backtesting, and FXMacroData timestamps each release so a test can ask what was known on the day.
- MCP servers and agents — tradingview-mcp runs nine preset strategies, MaverickMCP adds vectorbt tools, FinRobot wraps Backtrader, and AI Hedge Fund fills at the close with no costs modelled.
All 79 of them
Showing 79 of 79
Sort by
Backtesting Frameworks & Algo Trading Libraries21
A single-instrument Python backtester — one OHLC series, one strategy, no live trading.
An event-driven Python backtester with 122 indicators, frozen since April 2023.
Free hosted Python backtesting with bundled minute data and broker execution.
Python backtesting for allocation and rebalancing rules, not entries and exits.
A one-call wrapper over Backtrader, dormant since 2023 and broken on PyPI.
Gym-style market environments and deep-RL agents for trading research, MIT-licensed.
FinRL's successor, carrying one portfolio-weight vector from stock selection to Alpaca.
QuantConnect's engine without the cloud — your server, your data, your broker login.
One strategy class for backtest and live, plus a built-in LLM agent runtime.
Rust core, Python API, one strategy that backtests and trades live unchanged.
Python backtester with walk-forward model training and bootstrapped confidence bounds.
Microsoft's ML factor-research pipeline. The data its CLI downloads stops in late 2020.
An institutional-shaped Python backtester whose last commit to master was June 2024.
LEAN hosted for you — a browser IDE, mounted data and live trading through real brokers.
Zipline and Moonshot in Docker on your own hardware, wired to Interactive Brokers.
C#/.NET algo stack with 384 connectors; source is published but no longer open source.
Vectorised backtesting — thousands of parameter combinations in one NumPy pass.
Paid vectorbt — native Rust simulators, streaming indicators and a built-in MCP server.
Windows portfolio backtesting in C# or drag-and-drop blocks, with a built-in MCP service.
The maintained fork of Quantopian's Zipline. Bring your own data — it ships almost none.
Retirement & Portfolio Planning Calculators12
Backtest a portfolio of European UCITS funds on index data reaching back to 1970.
Run a withdrawal plan against every market cycle since 1871 and count how many survived.
Twelve withdrawal strategies replayed against every US market cycle since 1871.
The cFIREsim author's second simulator, with the taxes and accounts the first one lacks.
Any asset allocation, charted since 1970 in twelve countries' currencies and inflation.
Backtest an asset mix to 1972, run Monte Carlo, regress it on Fama-French factors.
Backtest an allocation on real fund prices and optimise it seven ways.
A US retirement model in an Excel workbook — taxes, Roth conversions, Monte Carlo.
The Pralana retirement model in a browser, with tax-form mock-ups and advisor seats.
Model a whole financial life as milestones and cash flows, then run it against history.
Backtest an allocation on daily data to 1885, using simulated pre-inception fund series.
A US retirement planner with linked accounts, Monte Carlo guardrails and Roth scenarios.
Stock Charting Platforms & Screeners8
Windows charting on ICE's consolidated feed, with EFS scripting and global exchanges.
Windows technical-analysis desktop for analysts, with Optuma's own EOD data bundled in.
Browser rebuild of Optuma, per seat, with scans and backtests run on Optuma's servers.
European charting and trading platform, free or paid for in euros or in trades.
Real-time US equity scanning on a Windows desktop, built from 500-plus alert conditions.
Charting across 50-plus markets, with Pine Script, server-side alerts and broker trading.
Automated technical analysis, real-time scanning and no-code backtesting in one web app.
Global fundamental screener with valuation models, screen backtests and an MCP server.
No-Code Strategy Builders & Auto-Trading7
Windows strategy generator — genetic programming writes the rules, you export the code.
Point-and-click strategy generation on Windows, exported as code to eight platforms.
No-code rule-based stock and ETF strategies, backtested and then traded for real.
No-code bots for defined-risk options spreads, run in the cloud through your own broker.
Multi-factor ranking, screening and point-in-time backtests without writing code.
Machine-searches for trading strategies, then tries to break them with robustness tests.
Block-built strategies for Indian and US markets, backtested per run, priced by live bot.
Desktop Trading Platforms & Order Flow5
Java desktop charting whose Elliott Wave tooling is the reason anyone buys it.
Windows charting, backtesting and order routing on EasyLanguage, sold outright or monthly.
Windows futures platform with C# strategy scripting and order-flow charting.
Windows order-flow terminal for 60+ connections, with an open C# API and a real free tier.
Windows-native charts, order flow and DOM trading, priced as packages plus exchange fees.
Options Data & Flow Analytics4
Options chains with raw IV, greeks and IVX surfaces over REST, US history to 2005.
Earnings-move statistics and option strategy screeners in a browser. No API.
Windows options backtester that replays positions on 5-minute US option history.
Smoothed options greeks and IV surfaces over REST, end-of-day back to 2007.
Trading Journal Software3
Trade journal with options spread detection on Elite and market replay down to 250ms.
The analytics-heavy journal: 150+ base charts across seven instrument types.
Auto-synced trade journal with bar-by-bar replay, backtesting and AI session review.
Fundamental Data & Stock Research Platforms3
Equibles
ClaimedFinancial research by browser, REST API and hosted MCP, with an AGPL self-hosted core
Thirty years of financials, a deep screener and 13F guru portfolios, sold by region.
Advisor research and client reporting, priced by quote rather than published.
Insider, 13F & Congressional Trade Trackers3
Short interest estimated daily from the lending pool, with cost to borrow beside it.
US political and SEC disclosure data — congress, insiders, lobbying — over one REST API.
Institutional 13F holdings back to 2001, queryable from a signed REST API.
AI Research Agents & Copilots3
LLM investor personas score a ticker, deterministic code sizes the book, nothing trades.
Open-source agents that turn a ticker and your own API keys into an analyst report.
A local finance research agent with 74 MCP tools, nine backtest engines and 14 brokers.
ETF Research & Screeners2
Dividend Trackers & Research2
A dividend calendar and forward-income projection wrapped around a portfolio tracker.
A US, Canadian and UK dividend tracker with broker auto-import, an MCP server and an API.
Stock Market Data APIs2
Financial MCP Servers2
Thirty-seven stock-analysis tools for any MCP client, on Yahoo data and no key.
Screeners, technical analysis and backtests for any MCP client. No TradingView account.
Macro & Economic Data APIs1
Release-timestamped macro data for 22 currencies, with a calendar and COT, from $50.
Market Analysis & Portfolio Optimization Libraries1
Portfolio optimisation as scikit-learn estimators — fit, predict, cross-validate.
FAQ
Can I backtest for free?
Yes, though every free plan cuts something different. Composer backtests stock and ETF rules for nothing and charges only for automated execution, testfolio keeps full available history on its free tier, and vectorbt and LEAN cost nothing but ship no usable market data. Portfolio Visualizer needs no login and caps ticker backtests at ten years, while TrendSpider and Trade Ideas offer no free backtest at all.
What is the difference between a backtest and a strategy tester?
Mostly where it lives. A strategy tester is the backtester built into a charting or trading platform, such as TradingView's for Pine Script or NinjaTrader's Strategy Analyzer for NinjaScript, and it tests on whatever history the plan lets the chart load. An engine such as LEAN or NautilusTrader is a library you call from code, with no chart, and the data, fill model and universe are yours to supply. Sierra Chart and eSignal test one chart at a time, which is the usual ceiling of a tester; AmiBroker and WealthLab test a whole portfolio.
Do I need intraday data to backtest?
Only if the strategy acts inside the day. A rule that rebalances daily or monthly runs correctly on daily bars, which is all Composer, Portfolio123 and the Optuma Web App use. A stop, a limit order or an opening-range entry needs finer bars, and that is where retail tiers draw their line — TrendSpider tests nothing finer than two-hour bars on its $82 plan and reaches one-minute at $183. Even then a bar is not the tape: vectorbt checks stops against OHLC, and Backtesting.py warns that it cannot know the path inside a candle.
Why do two backtests of the same strategy disagree?
Because each makes assumptions the other does not print. Composer fills at the daily close with one basis point of slippage, Backtesting.py at the next bar's open, and LEAN charges Interactive Brokers commissions with zero slippage; one runs on adjusted prices and another on unadjusted ones, where a split reads as a crash. The universe matters as much — a test that cannot see delisted stocks runs over the survivors, which is why Portfolio123 and WealthLab keep them. Adaptrade Builder's own FAQ lists mismatched date ranges and the MaxBarsBack setting among the reasons its numbers will not match your platform's.