forex-scalping.com

Limit order entries in an EA instead of market orders, what it did to fills and missed trades

my EA used to enter with market orders when a signal appeared. I looked at the fills and saw a consistent small cost from slippage and spread. So I tried switching to limit orders at the signal level, with an expiry of a few candles.

The effect was clear. Fills were at my price or better, the average cost per trade went down, and there was no slippage on the entry. But I also missed some trades, specifically the ones where price moved away without coming back. I wanted to see how those compared, so I logged the missed signals and what they would have made.

The missed trades were some of the best ones, which makes sense: strong moves don't pull back. But the ones I took with limits were better on average per trade, and the total difference was small. Which approach is better depends on the strategy, and the test showed it was a modest improvement for mine.

I kept the limit entries with a short expiry, and for a special class of signal that is momentum based, I still use market orders. It's a mix, depending on how the edge works.

I ran into the same tradeoff and tried a middle option: a limit order placed slightly beyond the signal level, so it fills on a small pullback instead of needing a full return. It kept most of the cost saving and caught more of the strong moves than a limit right at the level. Not all of them though. I still miss the ones that never pull back at all. I logged both versions on the same signals for a while before deciding, which is the only fair way to compare.

One technical thing to watch is the expiry. Some brokers do not support expiry on pending orders for every symbol or every account type, so the order just sits there. I handle it in the EA instead, by recording when the order was placed and deleting it after a set number of candles. I also cancel any pending order before high impact news and when the session filter closes. Otherwise a stale limit can fill hours later when the setup no longer exists.

On the missed trades being the best ones, I think it is worth accepting that as the price of the method rather than trying to fix it. When I tried adding a market order fallback for strong moves, the logic got complicated and the fallback trades had all the slippage I was trying to avoid. In the end I picked limits and accepted fewer trades. The important thing is that the backtest models the missed fills honestly, because a tester that fills every limit will overstate results.