Free Historical Data Sources for Forex & CFDs

The Sources, Honestly Rated

Source	Granularity	Spread info	Cost / friction	Honest caveats
Dukascopy historical feed	Tick (bid/ask)	Yes	Free; slow exports, format conversion	The reference free tick source; it’s one bank’s feed — see broker differences
TrueFX	Tick (bid/ask)	Yes	Free with registration	Majors-focused; monthly files to stitch
HistData.com	M1 OHLC	No	Free	Candles only — no spread, no intrabar path (what that costs you)
Your broker’s MT4/MT5 export	M1–D1 OHLC	No	Free, instant	Intraday history often shallow (months); mixed-source past; but it matches your live feed
cTrader backtesting data	Tick	Yes	Free with broker account	Tied to the platform; quality varies by broker
Stooq / Yahoo Finance	Daily	No	Free	Indices/stocks daily work only; fine for position-trading studies
Built-in replay-tool data	Per tool — check	Per tool	Free tier varies	Zero pipeline; you trust the tool’s sourcing — verify granularity and depth before relying on it

The Pipeline Tax (Budget for It)

The DIY route’s real cost is time, not money: download (Dukascopy tick exports run slow), convert (formats rarely match your tool), import, then patch — weekend stubs, missing minutes, server-clock alignment. Plan on hours before the first replay, and on repeating a slice of it whenever you extend the window. This tax is why most manual backtesters quietly downgrade to M1 candles — paying with granularity instead of hours, usually without noticing the trade.

The Zero-Pipeline Route

Replay tools with built-in data invert the deal: nothing to download, convert or patch — the data streams on demand. StrategyTune (built by this site’s team — disclosure) streams recorded bid/ask tick history for 70+ instruments in the browser, free, with per-instrument depth published in its live catalog; other tools bundle candle-level data of varying depth (tools comparison). The trade-off is control: you’re trusting the tool’s sourcing, so apply the same checks you’d apply to any feed — verify the granularity claim, run the Sunday-candle clock test, and for spread-sensitive strategies confirm the data carries real bid/ask rather than a synthetic constant spread.

Matching Source to Strategy

• Tight-stop intraday / scalping: tick data, full stop — Dukascopy/TrueFX via pipeline, or a tick-streaming replay tool. M1 candles are not enough; the test will guess your contested fills.
• Swing on H4/D1: M1–H1 candles suffice; broker exports or HistData are fine. Spend the saved hours on regime coverage.
• Position/portfolio studies: daily data from anywhere reputable; granularity is irrelevant at 80-pip stops.

Frequently Asked Questions

Is Dukascopy data representative if I trade with a different broker?

It's one Swiss bank's feed — high quality, but still a single source. For most strategies that's fine; for spread-sensitive or wick-precision systems, treat results as approximate until validated on data resembling your live broker, and apply the robustness checks from the broker-differences guide.

Why is my MT4 intraday history so short?

Brokers typically keep only recent intraday candles on their servers; deeper M1 history must be imported from external sources like Dukascopy or HistData. Note that imported history and live-collected candles can come from different feeds — your chart's past and present may not be the same data.

Are paid data vendors worth it for manual backtesting?

Usually not at retail scale — free tick sources plus a careful pipeline (or a replay tool with built-in tick data) cover manual backtesting needs. Paid vendors earn their fee for algorithmic research at scale: cleaned multi-venue data, guaranteed completeness, futures/options depth.

How do I verify a data source's quality quickly?

Four spot checks: count weekly daily-candles (five = standard clock), find a known news event and check its timestamp, compare a famous extreme (a flash-crash low) against references, and scan a random week for missing minutes. Twenty minutes of checks prevents weeks of backtesting on broken data.