Limit orders on ranges and the adverse selection that hides in them

[Amanhasnoname]

Placing limit orders at the edges of a range looks like the cleanest way to scalp. Buy at support, sell at resistance, collect the spread-like profit as price oscillates.

The catch is who gets filled. A limit order fills when price moves to it. If the market is about to break through that level, the order fills and price keeps going. If the level will hold, the order might miss by a tick as price bounces before reaching it.

This is adverse selection. The fills that happen are disproportionately the ones where price continues against the position, and the ones that would have worked often don't trigger.

In practice, limit entries show worse results than a backtest assuming fills whenever price touches the level. The bias is hard to remove without live testing.

Compensating measures include requiring price to trade through the level by a pip in simulations, and comparing fill rates between winners and losers in live data.

Does the adverse selection effect show up in your limit-order results?