Comparing execution on different pairs: slippage is not uniform

[Fairman]

Comparing execution on different pairs: slippage is not uniform

Slippage is often quoted as a single average figure, but it varies a lot by pair. EUR/USD on a quiet afternoon might fill at the quote nearly every time. GBP/JPY or an exotic cross at the same moment can slip a pip or more on identical order sizes.

The reason is depth. Majors have deep books across many providers, while crosses are often derived by combining two other pairs, which makes pricing more complicated. Some brokers even synthesize cross prices from major feeds, adding latency and markup.

Run an execution test per pair rather than assuming performance carries over. Use small, equal sizes, similar times of day, and enough trades for stable averages, at least 50 per pair.

Rank pairs by total execution cost relative to average movement. Often the pair with the smallest spread is not the one with the lowest overall friction.

Then restrict the trading universe to pairs that pass. A shorter watchlist of good fills beats a long one of marginal ones.

Which pair turned out to be the surprise winner or loser in your execution testing?

[Amanhasnoname]

My surprise loser was EURGBP, which I had assumed would be easy because its spread looked so small. After sixty trades with equal small size, my average slippage there was about 0.4 pips against 0.15 on EURUSD, and the stops filled worse in the quiet hours when the price jumped over my level. The headline spread was lower but the total cost per trade ended up higher.

The winner for me was USDJPY. The quoted spread was nothing special, but the fills were consistently at the price I saw, and over the same sample the cost relative to average M5 movement was the lowest of the group. I would never have guessed that from the spread alone.

I followed your idea of ranking by cost against average movement and cut my list from seven pairs to three. The first month after that, my trade count went down by a fifth and my net result improved slightly. Having fewer pairs also made it easier to learn how each one behaves around the open.