Concepts5 min read

Market Microstructure and Execution Shortfall

Market microstructure describes the mechanics by which orders become trades — the order book, queue priority, latency, and the discrete events that move price between quotes. Execution shortfall is the cost born from operating inside this machinery: the difference between the price at which a strategy decided to trade and the price it actually achieved. For systematic traders, ignoring shortfall means backtests print returns that the live account never sees.

The canonical decomposition comes from Perold (1988), who framed shortfall as the gap between a paper portfolio executing at decision prices and a real portfolio executing through a market.

Shortfall = (P_exec − P_decision) × Q_filled + P_decision × Q_unfilled × (P_close − P_decision) / P_decision + Fees

The first term is the realized execution cost on filled shares. The second is opportunity cost — what the unfilled portion would have earned or lost had it traded at the decision price. The third aggregates commissions, exchange fees, and rebates. All three components are denominated in the same currency unit, typically basis points of notional.

Interpreting the magnitude

Shortfall is asymmetric: it is positive when execution worked against the trader. For liquid US equities traded in modest size, total shortfall typically lands between 5 and 25 basis points round-trip, with the median strategy clustering near 10–15 bps. Futures contracts on deep instruments (ES, ZN, CL) can execute at 1–3 bps round-trip when crossing a single tick. Crypto perpetuals on major venues run 2–8 bps for taker orders, with maker rebates pulling some flows negative.

Above 30 bps round-trip on liquid instruments, the strategy is either trading too large for its time horizon, crossing the spread when it should be posting, or trading instruments where the modeled liquidity does not exist. A shortfall that exceeds the strategy's per-trade alpha is a strategy with no live edge, regardless of backtest Sharpe.

Decompose before optimizing. A strategy with 8 bps spread cost and 2 bps opportunity cost has a different fix than one with 2 bps spread cost and 8 bps opportunity cost. The first needs better order placement; the second needs faster signals or smaller size.

What shortfall does not capture

Shortfall measures cost relative to the decision price, but the decision price itself is a modeling choice. Using the last trade as decision price flatters strategies that signal off stale prints; using the mid at signal time penalizes them honestly. Backtests routinely use close-to-close or next-bar-open execution, which sidesteps shortfall entirely and produces results uncorrelated with live performance.

Shortfall also does not measure market impact on subsequent trades. A large parent order that walks the book leaves the market displaced; the next signal from the same strategy now faces a worse price that was caused by the prior trade. This shows up as autocorrelated execution cost, not as a shortfall term on any individual fill.

Finally, shortfall as defined above is silent on adverse selection. A maker order that fills consistently right before the price moves against it has zero spread cost — the maker captured the spread — but bleeds money on every fill. Adverse selection requires a separate post-trade markout analysis, measuring price drift over the seconds and minutes following each fill.

A backtest that uses bar-close fills with a fixed commission assumption is not modeling execution. It is modeling a frictionless market that does not exist. The gap between such a backtest and live trading is almost entirely shortfall, and it does not shrink with more data — it persists.

How Kestrel Signal presents it

Kestrel Signal computes execution shortfall on every simulated fill, separating the spread component, the opportunity cost on partial fills, and explicit fees. Decision prices are anchored to the bar at which the signal was generated, not the bar at which the order was filled, so timing slippage surfaces honestly. Post-trade markouts at 10s, 1m, and 5m horizons appear alongside shortfall to expose adverse selection that pure cost accounting hides. The platform reports shortfall in basis points of traded notional and as a percentage of gross strategy return, making it immediate whether execution is consuming the edge.