Average spread is a poor guide, median spread is better

[Fairman]

Average spread is a poor guide, median spread is better

Brokers advertise average spreads, and the figure is dominated by outliers. A few moments of 10 pip spreads at rollover lift the average noticeably, yet those moments do not matter if no trading happens then. Alternatively, a broker may show a low average computed only over liquid hours.

For scalpers, the median spread across the actual trading hours is far more representative. It shows what the typical entry costs, ignoring rare spikes. Then the tail, such as the 95th or 99th percentile, shows how bad things get.

Both numbers matter. A broker with a median of 0.2 and a 99th percentile of 3.0 is very different from one with a median of 0.4 and a 99th percentile of 0.9. The first looks cheaper but delivers occasional painful fills. The second is steadier.

Collect tick data for the traded window and compute the full distribution. A histogram makes differences between brokers instantly visible.

Then decide whether tight and spiky or slightly wider and steady suits the strategy better.

Which spread statistic has proved most useful when comparing brokers?