Two tick sources, two different backtests
Posted: Sat Oct 31, 2026 8:19 am
Download tick history for the same pair and the same week from two different providers and overlay them. They will not match. Tick counts differ, timestamps drift by tens or hundreds of milliseconds, and a news spike may appear on one feed and be smoothed on the other. Each provider aggregates different liquidity sources, and some record every top-of-book change while others filter small ones.
For a swing system that is background noise. For a scalper working 2 to 5 pip targets it can flip the verdict, because the whole edge may live in micro-moves around a level that one feed shows as a touch and another does not.
So run the identical system on both sources and compare trade count, win rate and average trade. If those move a lot, the edge depends on the data, which usually means it is fragile. If they stay close, that tells more than any quality percentage.
Keep in mind that neither dataset is the broker's own stream. The only tick record that truly represents live fills is the one the trading platform logged itself, and that is worth the disk space.
Have you compared two feeds on the same system, and how far apart did the trade counts land?
For a swing system that is background noise. For a scalper working 2 to 5 pip targets it can flip the verdict, because the whole edge may live in micro-moves around a level that one feed shows as a touch and another does not.
So run the identical system on both sources and compare trade count, win rate and average trade. If those move a lot, the edge depends on the data, which usually means it is fragile. If they stay close, that tells more than any quality percentage.
Keep in mind that neither dataset is the broker's own stream. The only tick record that truly represents live fills is the one the trading platform logged itself, and that is worth the disk space.
Have you compared two feeds on the same system, and how far apart did the trade counts land?