ADV Participation Limits in Research

A backtest can look profitable on paper while quietly assuming trades no real desk could execute — for instance, buying an amount equal to half a day's entire trading volume in one name without moving the price. ADV participation limits are a research-stage guardrail: before a backtest is trusted, every simulated trade is checked against a cap, typically 5-10% of that stock's average daily volume (ADV), and any trade exceeding the cap is either scaled down or spread across multiple days in the simulation.

The logic is straightforward: if your real capital, divided across signals, would need to trade more than that fraction of ADV in a name to hit the target position, the backtest's assumed fills are unrealistic and its returns are overstated relative to what could actually be traded at scale. Applying the cap consistently across the whole backtest also reveals how a strategy's capacity shrinks as the fund gets bigger — a signal that looks great at $50 million might already be breaching participation limits at $500 million, and the backtest should show that degradation rather than hide it.

In practice this check is applied per name, per day, using either static historical ADV or a rolling estimate, and strategies are often reported both with and without the cap so researchers can see exactly how much of the apparent edge depended on unrealistic execution assumptions.