A limit order 2 pips better than market: why the fills you get are the worse ones

[Fairman]

A limit order 2 pips better than market: why the fills you get are the worse ones

A signal fires with EURUSD at 1.0852. You can buy at market, or put a buy limit at 1.0850 and save 2 pips. The stop is 1.0846 either way, the target 1.0861.

Market entry: risk 6, reward 9. Limit entry: risk 4, reward 11. That looks obviously better, but there's a catch in who gets filled.

Let's use 100 signals and make up round numbers. At market, 45 reach the target and 55 hit the stop, so the result is 45 times 9 minus 55 times 6, which is +75 pips before costs.

Every trade that hits the 1.0846 stop has to trade through 1.0850 on the way down, so the limit fills on all 55 losers. Winners are different: some run straight away and never pull back 2 pips. Suppose 20 of the 45 winners never touch 1.0850. The limit fills 25 winners at +11 and 55 losers at -4. That's 275 minus 220, which is +55 pips. Worse than the +75 at market, on fewer trades.

The break-even is the share of winners missed. Limit result is 45 times (1 - m) times 11 minus 220, and it equals 75 when m is 40.4%. If more than about 40% of your winners run without a pullback, the limit lowers the result.

The loser fill rate is 100% and the winner fill rate is lower. That's the selection problem, and the entry type alone won't remove it.

[FXS Prop Desk]

Fairman, the limit order example captures adverse selection well: the fill comes mostly when price is moving against you. A way to measure it on your own data is to compare, for filled limit orders, how often the stop was hit against market entries taken on the same signals. If limit fills hit the stop noticeably more often, the 2 pips saved is being paid back elsewhere.