Testing on one broker's data and trading on another's
Posted: Sat Oct 31, 2026 10:12 pm
A strategy developed on the tick data from Broker A is deployed on Broker B. The logic is unchanged, but the results shift.
Differences accumulate. Spreads vary by a few tenths of a pip. Quote timing differs, so entries trigger at different moments. Spike filters differ, so some wicks exist on one feed and not the other. Rollover handling differs, and so does the daily bar cut-off.
For a strategy with a 0.5 pip edge, those differences are as large as the edge itself. A system that was profitable on A can lose on B simply because the feeds differ.
Testing on several brokers' data is a robustness check. If the strategy only works on one feed, it's probably exploiting something peculiar to that feed, such as a lag or a specific filter. That's a fragile edge.
If it works on several, the edge is more likely related to the market itself.
A good practice is to record live ticks from the actual trading broker and run a final validation on that data before going live with size.
How much did the results change when a strategy was moved from one broker's data to another's?
Differences accumulate. Spreads vary by a few tenths of a pip. Quote timing differs, so entries trigger at different moments. Spike filters differ, so some wicks exist on one feed and not the other. Rollover handling differs, and so does the daily bar cut-off.
For a strategy with a 0.5 pip edge, those differences are as large as the edge itself. A system that was profitable on A can lose on B simply because the feeds differ.
Testing on several brokers' data is a robustness check. If the strategy only works on one feed, it's probably exploiting something peculiar to that feed, such as a lag or a specific filter. That's a fragile edge.
If it works on several, the edge is more likely related to the market itself.
A good practice is to record live ticks from the actual trading broker and run a final validation on that data before going live with size.
How much did the results change when a strategy was moved from one broker's data to another's?