Academic Cost Assumptions vs Real Costs

Academic factor papers commonly assume a fixed cost per trade — say, 10 basis points round-trip, applied uniformly regardless of stock size, trade size, or market conditions. That number is easy to plug into a backtest and easy to defend in a footnote, but it's rarely close to what a real fund pays. Real costs scale with how much of a stock's daily volume you're trying to trade (market impact rises faster than linearly as your order size grows), spike during volatile periods exactly when a strategy might want to trade most, and are dramatically higher for small-cap and micro-cap names than for large, liquid ones — precisely the stocks where many published anomalies (value, momentum applied to small caps) show their strongest paper returns.

Worked example

A published paper reports a small-cap value strategy earning 8% annualized alpha before costs, using a flat 10 bps round-trip assumption that shaves off roughly 1% a year, leaving 7%. A trading desk replicating the strategy at real scale finds that trading illiquid small caps at the required turnover actually costs closer to 80–150 bps round-trip once market impact is included, because the position sizes needed to matter push well past what the stock's daily volume can quietly absorb. That turns the same 8% pre-cost alpha into a return close to zero or negative once capacity-adjusted implementation costs are applied.

This mismatch is a major reason capacity — how much money a strategy can actually run before its own trading erodes the edge — is treated separately from a paper's headline Sharpe ratio.