IOC Limit Fills and Selection Bias in Momentum Futures Trading

Summary

The document describes execution concerns for an XGBoost signal model trading MNQ micro futures, with momentum among its features and several signals per day. The author reports that market-order slippage averaged 1.07 points and says this erased the strategy’s edge, then describes switching to immediate-or-cancel limit orders with a 0.50-point offset. They are tracking daily fills and rejections.

The key concern is that fast, successful momentum moves may leave the limit price before an order fills, while setups that retrace may be more likely to execute. If so, the filled trades could represent a different subset of signals and alter the strategy’s exposure to momentum versus reversal. The document asks whether this selection effect is known, what fill rate to expect, and what execution alternatives might control slippage. It provides no fill-rate evidence or tested comparison, so it does not resolve those questions.

Key ideas

• Market-order slippage is reported as a major drag on the futures strategy’s edge.
• The author switched to IOC limit orders with a 0.50-point offset and tracks fills versus rejections.
• Fast moves that follow momentum signals may be more likely to miss the limit order.
• The resulting filled trades could overrepresent setups that retrace toward the limit price.
• No expected fill rate or alternative execution method is established.

Tags

# IOC Limit Orders in momentum-based futures strategy – selection bias and fill rate?


# IOC Limit Orders in momentum-based futures strategy – selection bias and fill rate?












I'm running an XGBoost-based signal model on MNQ (Nasdaq micro futures) generating 5-10 signals per day. The model uses momentum as one of its key features.

Problem: With market orders, average slippage was 1.07 pts, which destroyed the edge entirely (first live day: -$71, mostly slippage).

What I tried: Switched to IOC Limit orders with 0.50 pt offset.

My concern: This may introduce a selection bias:

- Momentum signals that "work" move quickly away from the entry level → IOC gets rejected

- We only fill setups where price comes back → mean-reversion bias

- This fundamentally changes what the strategy is actually trading

Data I'm collecting: Fill rate (fills vs rejects per day) – currently tracking ioc_fills_today / ioc_rejects_today.

Questions:

- Is this selection bias a known issue with IOC limits on momentum strategies?

- What fill rate is realistically expected for IOC with 0.50 pt offset on NQ?

- Is there a better execution approach for momentum futures that controls slippage without this bias?

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.