Backtests built on bad tick data
Posted: Fri Oct 30, 2026 7:57 am
A backtest is only as honest as the data under it. Many retail platforms generate synthetic ticks from one-minute bars, interpolating a path that price may never have taken. For a scalper with a 3 pip target and a 3 pip stop, the order in which price touched the high and low inside a candle decides whether the trade won or lost, and interpolation guesses it.
The result is a test that looks precise and isn't. Equity curves come out smooth, win rates look healthy, and live trading disappoints.
Better data helps. Real tick data from a reputable source, ideally from the same broker that will fill the orders, captures actual spread behaviour and the true path of price. It's bigger and slower to process, but the difference in results can be dramatic.
Even then, quality varies. Gaps, bad ticks and missing periods corrupt tests quietly, so cleaning and sanity checks are part of the job.
What data source feeds the backtests, and has a live month ever been compared against a test of the same month?
The result is a test that looks precise and isn't. Equity curves come out smooth, win rates look healthy, and live trading disappoints.
Better data helps. Real tick data from a reputable source, ideally from the same broker that will fill the orders, captures actual spread behaviour and the true path of price. It's bigger and slower to process, but the difference in results can be dramatic.
Even then, quality varies. Gaps, bad ticks and missing periods corrupt tests quietly, so cleaning and sanity checks are part of the job.
What data source feeds the backtests, and has a live month ever been compared against a test of the same month?