Spread dynamics and session selection

Use observed spread and activity to decide when a strategy should trade, rather than relying on a fixed clock alone.

Spread is not a constant input

A fixed spread assumption can hide when a strategy's edge is consumed by transaction cost. Measure spread distributions by session, day, symbol, and volatility condition rather than using one average for every trade.

A session with more movement is not automatically better if the spread and slippage rise faster than the opportunity.

Build a session hypothesis

A session filter should have a reason: more liquidity, a defined opening behaviour, a particular overlap, or a known execution constraint. Test the filter with the same strategy and costs, not only by selecting the session with the best historical curve.

Use broker server time consistently.
Compare gross and net results by session.
Include trades that start near a session boundary.
Review whether the filter simply removed a difficult regime after the fact.

Make abnormal spread a safety state

An EA can refuse new entries when spread exceeds a documented threshold, while still managing existing risk. The threshold needs a safe fallback when price data is unavailable and should be monitored so it does not silently block the system for weeks.

Execution controls should reduce avoidable cost without becoming an untested source of look-ahead.

Discuss this article

Ask a setup question, share a backtest, or compare notes with other algorithmic traders. Use a display name; your email remains private.

Create an account or sign in above to join the conversation.

Loading discussion…