Order books, queue position, and why limit order scalping is a game of timing

[Fairman]

Order books, queue position, and why limit order scalping is a game of timing

On venues with a visible queue, the first order at a price gets filled first. In forex pools run by ECNs or matching engines, similar priority rules apply. A limit order placed a split second late sits behind the early ones and fills only after they do.

That structure changes the economics of passive scalping. Fast participants get the best queue positions and capture most of the profitable fills. Slower participants get filled mainly when the level is about to break, meaning the fill comes with bad news attached.

For retail scalpers with ordinary latency, the practical takeaway is to avoid competing for the touch in busy markets. Instead, place limits where fast participants have less interest, such as slightly behind the obvious level, or on pairs and sessions with less algorithmic crowding.

Another approach is accepting the spread and using market orders when the setup is strong, since the cost of waiting in a queue can be larger than the cost of paying the spread.

How has passive order placement performed compared with simply crossing the spread?

[Fairman]

I tried placing limits one pip behind the obvious level instead of right at it. Example: range low at 1.0850, my buy limit at 1.0849 instead of 1.0851. Fill rate dropped, roughly 40 percent of signals filled instead of 65 percent. But the filled trades won more often, about 60 percent versus 52. The overall result was slightly better per signal, and the trades felt less like I was getting run over. Not a huge improvement, but consistent over about 300 signals. Small changes in placement can matter more than expected.

[Amanhasnoname]

I tested the passive approach on EURUSD over about 250 signals last year, and my experience was a bit different from a neat story. Limits at the touch filled about 62 percent of the time, but a good share of those fills came right before the level broke, and the average trade was around 0.4 pips worse than the signals that never filled. Crossing the spread on strong setups had lower variance and, to my surprise, a similar net result.

What I do now is split the decision by setup quality. For the A setups, with trend, level and momentum aligned, I pay the spread and use a market order, because waiting in a queue costs more than the 0.3 pips. For the B setups, I place the limit about a pip behind the level and accept the lower fill rate.

I also stopped placing orders in the first minute after a major release, since the queue there is dominated by faster participants and my fills in that window were consistently the worst.