A practical A/B test protocol for two brokers
Posted: Thu Oct 22, 2026 6:09 am
Comparing brokers based on reviews and marketing is unreliable. A direct comparison with your own trading gives real answers.
Protocol:
1. Open accounts at both brokers with similar balances.
2. Use the same platform (e.g. MT5) and the same VPS if possible.
3. Trade identical signals on both, at the same size. An EA that copies to both helps.
4. Run for at least 4 weeks or 100 trades.
5. Record: spread at entry, entry fill, exit fill, slippage, commission, rejects.
Analysis:
- Compare total cost per trade.
- Compare slippage distributions, especially tails.
- Compare results during news and quiet periods separately.
- Compare reject rates.
Pitfalls:
- Small samples can be misleading. 20 trades isn't enough.
- Different server times may cause signals to differ slightly.
- One broker might have an outage during the test, skewing results.
Has anyone run a structured broker comparison like this? What did you find?
Protocol:
1. Open accounts at both brokers with similar balances.
2. Use the same platform (e.g. MT5) and the same VPS if possible.
3. Trade identical signals on both, at the same size. An EA that copies to both helps.
4. Run for at least 4 weeks or 100 trades.
5. Record: spread at entry, entry fill, exit fill, slippage, commission, rejects.
Analysis:
- Compare total cost per trade.
- Compare slippage distributions, especially tails.
- Compare results during news and quiet periods separately.
- Compare reject rates.
Pitfalls:
- Small samples can be misleading. 20 trades isn't enough.
- Different server times may cause signals to differ slightly.
- One broker might have an outage during the test, skewing results.
Has anyone run a structured broker comparison like this? What did you find?