Is Yahoo Finance forex data good enough for backtesting?

Daily, usually yes. Intraday, no. Measured on EURUSD, its 15-minute bars carried 15 distinct closes across 40 bars.

How do I check if my price data is stretched?

Distinct closes divided by bar count. Real intraday on a liquid name is often 60 percent distinct or higher. Under about 30 percent means values are repeating.

Can I backtest forex without paying for data?

Yes on daily bars, or by exporting your own broker history. That export includes your actual spread.

Why is free intraday forex data unreliable for backtesting?. Kareet.

Why is free intraday forex data unreliable for backtesting?

2026-07-294 min

Short answer

Most free forex feeds have no true intraday grain. They repeat hourly values across every minute or 15-minute slot. The chart looks normal. Those bars never traded.

The short answer

Forex has no central exchange. Every price is one venue's quote.

Free feeds often hold only hourly resolution. Asked for 15-minute bars, they repeat the hourly value four times. The chart looks fine. No such 15-minute bar ever traded.

How to spot it

Count distinct closing prices as a fraction of bar count.

Real minute data on a liquid pair is mostly distinct closes. Stretched data repeats.

A common free feed, 40 EURUSD 15-minute bars: 15 distinct closes. Values repeating exactly four times each. That is one hourly print stretched across an hour.

A real feed on the same pair and timeframe: 66 distinct closes out of 76.

Why it matters

A backtest on stretched data does not fail. It prints a clean curve. Nothing in the output says the input was fake.

The fake data is smoother than life. Gaps vanish. Wicks vanish. Stops that would have hit do not. Small-move rules look better than they are.

What is actually usable

Your broker export. Closest to what you would have traded, including your spread.
A paid feed with real tick or minute aggregation.
Daily bars from almost anywhere. Daily usually has real daily resolution behind it.

The check

Before you trust a test, count distinct closes on the data that fed it. One line. It will occasionally save you months of confidence in a result that was never real.

Questions

Is Yahoo Finance forex data good enough for backtesting?: Daily, usually yes. Intraday, no. Measured on EURUSD, its 15-minute bars carried 15 distinct closes across 40 bars.
How do I check if my price data is stretched?: Distinct closes divided by bar count. Real intraday on a liquid name is often 60 percent distinct or higher. Under about 30 percent means values are repeating.
Can I backtest forex without paying for data?: Yes on daily bars, or by exporting your own broker history. That export includes your actual spread.