This review is written by the team that builds FXAbsolute, which makes it biased by construction. Read it sceptically. To make it worth anything, the real limitations come first — including one that can lose months of your work.
Every session, trade and journal note is stored locally in your browser. Nothing is on a server. That means it is private and fast, and it also means clearing your browser data deletes everything. Switch from your phone to your laptop and none of your history follows. If you have built up fifty sessions of work, that is a real exposure, and today there is no way to back it up.
This is the most serious flaw in the product. It is a consequence of the design that also makes it free and account-free, but that does not make it less painful when it bites.
Nine forex pairs, gold, three indices and two crypto. No equities, no exotic pairs, no commodities beyond gold. If you trade USDTRY or Tesla, this tool has nothing for you.
The archive is mid-price OHLC. Two consequences worth internalising: any volume-based method — VWAP, volume profile, OBV — cannot be tested here at all; and any strategy targeting a few pips will look better than it would live, because the spread you would actually pay is absent. A scalping result on this platform is an optimistic result.
You cannot write a strategy in code and run it over ten years unattended. Every trade is placed by hand, bar by bar. That is deliberate and it is the point of the tool, but if you want automated parameter sweeps, this is the wrong instrument and something like QuantConnect is a better fit.
Coverage runs to 2026 and there are no live quotes. This is a practice and research environment, not a trading terminal.
Real ones. The largest is an eight-day hole in GBPUSD in February 2021, with several three-to-five day gaps elsewhere. Every free dataset has these; the difference is that this one publishes the list.
No account, no card, no trial period, no locked instruments, no export limits. The comparison worth making is against paid alternatives where the free tier is a demo — here the whole tool is the free tier. You can be replaying real candles about ten seconds after landing on the site.
28,412,683 one-minute bars spanning 2021 to 2026, pre-assembled. Free tiers elsewhere tend to cap requests hard — Alpha Vantage allows 25 per day, which cannot realistically deliver five years of M1 — or hand you raw tick archives you must stitch together yourself. The assembly work is genuinely tedious and it is where most people's data projects quietly die.
Bar-by-bar advance with adjustable speed, drawing tools, limit orders, stops and targets, an automatic trade journal, and statistics computed as you go — win rate, profit factor, average win and loss, total pips. The journal filling itself matters more than it sounds, because the main reason traders stop journalling is the friction of doing it by hand.
An MCP server exposes the whole archive to Claude Code, Cursor or any MCP client, and can pair with a live chart so an agent sees what you see. Other market-data MCP servers exist — Alpha Vantage, Polygon, TraderMade among them — but none of them drive a trader's chart. If you work in a terminal, this is the differentiator.
Browser-based with no install, and it works on a phone properly rather than nominally. For anyone whose main computer is a phone, that is not a small thing.
| If you are… | Verdict |
|---|---|
| Learning price action and needing screen time | Strong fit. Free, fast, and the journal builds the habit for you. |
| Preparing for a prop firm challenge | Strong fit. Rehearse the rules on real data at no cost. |
| Researching statistics over years of bars | Strong fit, especially through the agent bridge. |
| Trading volume-based methods | Wrong tool. There is no volume data. |
| Backtesting equities or exotic pairs | Wrong tool. Fifteen instruments, none of them stocks. |
| Running automated parameter optimisation | Wrong tool. Manual execution only. |
| Scalping with a two-pip target | Use with caution. No spread means results flatter you. |
The honest summary: paid tools buy you breadth, automation and tick-level accuracy. This buys you depth of history and zero friction.
Forex Tester and Soft4FX both offer more instruments, tick-level modelling and automation, and both cost money and require installation — Soft4FX also requires MetaTrader. TradingView's bar replay is excellent but its useful depth sits behind a subscription. If tick accuracy or automation is central to your method, pay for one of those.
If what you need is thousands of repetitions on real price action to build judgement, the free option is not a compromise — repetition count is what matters, and nothing here rations it.
FXAbsolute is a focused tool that does one thing well: manual bar-by-bar practice on real historical data, free, with no barrier to starting. It is genuinely useful for building execution skill and for statistical research through the agent bridge.
It is not a complete trading platform, and it should not be your only tool. The instrument list is narrow, the absence of volume and spread rules out whole categories of strategy, and — most importantly — your work lives in one browser and can be lost. Until that changes, export anything you cannot bear to lose.
Judge it by whether it increases the number of decisions you make on real price action per week. On that measure it does well. On most other measures, it is deliberately narrow.
Yes, entirely. No account, no payment details, no trial period, and no locked instruments. All 15 instruments and the full five years of one-minute data are available without signing in.
It is a real, working browser-based backtesting tool with a public MIT-licensed MCP server on GitHub. It takes no payment details and requires no account, so there is no billing risk. Its main practical risk is that session history is stored only in your browser and can be lost if you clear browser data.
Session history lives only in your browser and does not sync between devices, so clearing browser data deletes it. It also covers only 15 instruments, holds no volume data and no bid/ask spread, supports manual testing only with no coded automation, and provides no real-time quotes.
No. The archive is mid-price OHLC only. Forex has no central exchange and therefore no true market-wide volume, so the field is omitted rather than filled with a feed-specific tick count that could mislead.
Yes. It runs in a mobile browser with no installation, and the interface is built for touch rather than merely tolerating it.
Forex Tester offers more instruments, tick-level modelling and automation, and costs money with a desktop installation. FXAbsolute is free, browser-based and manual-only with 15 instruments. Choose Forex Tester for tick accuracy or automation; choose FXAbsolute for free repetitions on real data with no setup.
No account, no download. Open a chart and start replaying real candles — then decide whether this review was fair.
Open FXAbsolute →