AmiBroker review (2026)

AmiBroker remains, in 2026, the reference desktop backtester for retail systematic traders. A one-time $299-$499 purchase buys an engine that outruns most subscription cloud platforms on raw testing speed. Its core loop is unmatched at the price: vectorized AFL signal generation feeding a bar-by-bar portfolio simulator, with walk-forward optimization, Monte Carlo simulation, and a custom backtester interface for anything nonstandard. Version 7.00.1 (December 2025) shows the product is still actively developed, and the 24-month free-upgrade policy plus roughly half-price upgrades afterwards keep total cost of ownership low.

The trade-offs are structural rather than incidental. AmiBroker sells you an engine, and the ecosystem around it is your job: no data is bundled, live trading is a low-level DIY interface to Interactive Brokers that the vendor itself says should not run unsupervised, the platform is Windows-native only, and AFL is a proprietary language whose fluency transfers nowhere else. Users who budget time (and often a Norgate data subscription) for the setup phase report being very satisfied for a decade or more. Users expecting a turnkey TradingView-style experience tend to bounce off within the trial.

Our verdict: the best capability-per-dollar in desktop backtesting for equities, futures, and forex portfolios, provided you accept the data plumbing, the Windows dependency, and the fact that execution automation is a project rather than a feature.

AmiBroker is a Windows desktop application for technical analysis, charting, scanning and, above all, portfolio-level backtesting and optimization. That last item is its real claim to fame. Developed continuously since the late 1990s by a small Polish company (Tomasz Janeczko's AmiBroker.com), it is sold as a perpetual license rather than a subscription: $299 for Standard, $379 for Professional, $499 for the Ultimate Pack Pro bundle, each with 24 months of free upgrades. The current licensed release is 7.00.1 (December 2025), and a 30-day free trial is available without signup.

Strategies are written in AFL (AmiBroker Formula Language), an array-oriented scripting language purpose-built for trading logic. A moving-average crossover is one line. A full rotational momentum system over a few thousand symbols is a short script. Around the engine sit charting, real-time quote windows (feed-dependent), an exploration and scanner tool, and two companion apps: AmiQuote for downloading data, and the AFL Code Wizard, whose AI assistant generates code from plain English.

This is a power tool for individual quants, not a packaged retail product. There is no cloud component, no social features, no hand-holding. The learning investment is real, but the ceiling sits far above chart-platform backtesters like TradingView's.

The engine is a hybrid design. AFL evaluates indicators and signals as vectorized array operations, which is very fast, and the portfolio backtester then simulates bar by bar across the whole watchlist, handling capital allocation, position sizing, ranking, and margin. The Professional edition parallelizes analysis across up to 32 threads. The vendor's 'world's fastest portfolio backtesting' marketing is, unusually, broadly echoed by user benchmarks: full-market EOD tests over decades of data routinely finish in seconds.

Accuracy tooling is a genuine strength. Walk-forward optimization is built in, as are Monte Carlo simulation of equity curves, Maximum Adverse/Favorable Excursion charts (Professional), configurable commissions, slippage handling via trade delays and price arrays, and the detrending safeguards the manual actively encourages. The Custom Backtester Interface exposes the simulation loop at three levels of depth, so users can implement scaling, complex portfolio constraints, or custom metrics. That is the feature advanced users cite most when explaining why they stay.

The limits are worth stating plainly. Intrabar realism depends on your data granularity, and tick or second bars require Professional plus a capable feed. There is no native options or multi-leg simulation. And as with any flexible engine, look-ahead bias is the user's responsibility: future-referencing functions will happily produce fantasy results if misused.

AmiBroker ships with no market data, and this is the single biggest source of friction for new users. The database is local: you populate it via the AmiQuote downloader, an ASCII/CSV importer, or a real-time plugin. AmiQuote (bundled in the Ultimate Pack, $99 standalone with lifetime upgrades) pulls free EOD data from Yahoo Finance and Stooq, limited free intraday from Barchart OnDemand, and paid sources such as Tiingo.

Free data is fine for learning, but it has known problems for serious research. Yahoo's adjustments are occasionally wrong, coverage of delisted symbols is absent, and survivorship bias will flatter any long-only equity backtest. The community's standard answer is Norgate Data, a separate subscription that integrates tightly with AmiBroker and includes delisted US and Australian equities. That pairing is so common it has become the platform's de facto data layer for stocks. Real-time users connect eSignal, IQFeed, or Interactive Brokers feeds; tick charts require Professional and one of these sources.

Budget accordingly. A realistic serious setup is the $379 Professional license plus a data subscription of roughly $10-$100+/month, depending on asset class and resolution. Our data_quality score reflects what ships in the box (nothing), tempered by the strong vendor ecosystem around it.

Verified on the official order page in August 2026:

Item	Price	Notes
Standard Edition	$299	32-bit only, 10 RT symbols, 2 threads, minute bars max
Professional Edition	$379	64-bit, unlimited RT symbols, tick/second bars, 32 threads
Ultimate Pack Pro	$499	Professional + AmiQuote + AFL Code Wizard/AI (saves 10%)
Standard upgrade	$159	Only after free-upgrade period ends
Professional upgrade	$189	Only after free-upgrade period ends
AmiQuote alone	$99	Lifetime upgrades
AFL Code Wizard alone	$99	Lifetime upgrades

Every license includes 24 months of free upgrades and support, is licensed per person (installable on multiple machines you own), and never expires. Prices rose from the long-standing $279/$339, which is why many third-party reviews still show the old numbers.

Against the field, three years of AmiBroker Professional costs $379 in total, plus $189 if you want to stay current afterwards, versus roughly $2,000+ for three years of comparable cloud or charting subscriptions. The honest caveat is that data costs extra, and most serious users add a subscription that will eventually exceed the price of the software itself. Even so, on capability per dollar this is arguably the best deal in desktop backtesting.

AmiBroker can trade live, but this is its weakest pillar and buyers should be clear-eyed about it. The official path is IBController, a free bridge to Interactive Brokers' TWS invoked from AFL (GetTradingInterface). It supports market, limit, stop and trailing orders, brackets, OCA groups, account and position queries, and FA allocations. That is enough to run genuine automated systems, and many users do.

The vendor's own documentation is unusually candid, though: the interface is 'low-level', in long-running beta because of IB API churn, mandates paper-trading first, and 'is NOT intended to be run without supervision.' There is no built-in strategy-state management, no order-recovery logic, no multi-broker abstraction. You write that in AFL yourself, or you buy third-party bridges, which exist for various regional brokers but are unofficial. Nor is there any cloud deployment: your Windows machine, or a VPS, has to stay up.

Compare NinjaTrader or QuantConnect, where execution is a first-class product feature. The pattern in user feedback is consistent: research and signal generation in AmiBroker is world-class, while execution is a self-built project bolted on. Traders who want research-to-live in one supported pipeline should look at those alternatives. Traders who place orders manually from AmiBroker's exploration output, a very common workflow for EOD systems, lose nothing.

AmiBroker fits a specific and large niche: the self-directed systematic trader, usually on EOD or swing timeframes in equities, futures or forex, who wants institutional-grade portfolio testing without institutional pricing or a full Python stack. If you run ranking-based rotational systems, market-wide scans, or parameter optimizations with walk-forward validation, it is hard to find more capability for a one-time $379.

It is a poor fit if you are on a Mac and unwilling to run Wine or a VM; if you want bundled, clean data out of the box; if you need supported hands-off live execution across brokers; if you trade options strategies; or if you want transferable skills. Time invested in AFL pays off only inside AmiBroker, which is why Python-first quants usually choose backtrader, vectorbt, or QuantConnect instead, despite the extra setup.

The typical happy user in community threads bought a license years ago, pairs it with Norgate data, runs nightly explorations, and considers the software effectively free by now. The typical unhappy user expected TradingView-style immediacy and hit the data-setup wall in week one. The 30-day trial (no signup) is genuinely enough to find out which one you are. Commit to importing real data and writing a few AFL systems during it, though, or you will have evaluated the installer rather than the platform.