Benchmark a new instrument before it gets real size
Posted: Sat Oct 31, 2026 8:47 am
Adding a pair or index because it moved nicely last week is a classic way to import costs nobody measured. A short benchmark before any real trade protects against that.
Collect about a week of data with the platform running and no trades placed. Record the average and the 90th percentile spread by hour, the typical one-minute and five-minute range, and how often spreads blow out near the open and the rollover. Then compute the ratio of average range to average spread. A rough rule many scalpers use is wanting the five-minute range to be at least 10 to 15 times the spread, though the right figure depends on the strategy and is worth deriving from the pairs already traded.
Next, place a batch of micro-lot trades purely to measure slippage and fill time, as with any new broker. Compare the cost per trade against the existing pairs. If it is double, the same setup needs twice the edge to break even.
Only after that, run the strategy at minimum size for a set number of trades. The instrument earns its place by the numbers, not by a single good session.
What range-to-spread ratio do you demand before adding a new pair?
Collect about a week of data with the platform running and no trades placed. Record the average and the 90th percentile spread by hour, the typical one-minute and five-minute range, and how often spreads blow out near the open and the rollover. Then compute the ratio of average range to average spread. A rough rule many scalpers use is wanting the five-minute range to be at least 10 to 15 times the spread, though the right figure depends on the strategy and is worth deriving from the pairs already traded.
Next, place a batch of micro-lot trades purely to measure slippage and fill time, as with any new broker. Compare the cost per trade against the existing pairs. If it is double, the same setup needs twice the edge to break even.
Only after that, run the strategy at minimum size for a set number of trades. The instrument earns its place by the numbers, not by a single good session.
What range-to-spread ratio do you demand before adding a new pair?