Testing on one broker's data and trading on another
Posted: Sat Oct 31, 2026 7:33 am
Each broker's price feed differs slightly. Spreads, candle highs and lows, and the timing of ticks vary. A strategy tuned on one feed may behave differently on another.
For scalping, the differences are enough to matter. A stop that was never hit on the test feed may have been touched on the trading one. A setup that appeared clean might have looked messy.
Testing across two or three different feeds is a good robustness check. If the strategy survives on all, the edge is probably not an artefact of one data set. If it only works on one, the result is suspect.
Ideally, testing uses data from the broker that will actually be used. If that is not possible, a third-party source with realistic spreads is the next best choice.
Out-of-sample testing on a feed the optimisation never touched adds another layer of protection.
Has the same strategy been tested on different brokers, and how closely did the results match?
For scalping, the differences are enough to matter. A stop that was never hit on the test feed may have been touched on the trading one. A setup that appeared clean might have looked messy.
Testing across two or three different feeds is a good robustness check. If the strategy survives on all, the edge is probably not an artefact of one data set. If it only works on one, the result is suspect.
Ideally, testing uses data from the broker that will actually be used. If that is not possible, a third-party source with realistic spreads is the next best choice.
Out-of-sample testing on a feed the optimisation never touched adds another layer of protection.
Has the same strategy been tested on different brokers, and how closely did the results match?