Trading Into Alpha: Execution With A Signal

Standard execution algorithms are built on a convenient but limiting assumption: that the trader has no information about where the price is headed next, so the only goal is minimizing cost and risk against a neutral market. That assumption breaks the moment a trader is executing because they believe the price is about to move — a portfolio manager acting on a short-term signal isn't indifferent to speed, because every minute of delay risks losing part of the very edge the trade is meant to capture.

Why a signal changes the optimal schedule

Without a signal, the classic trade-off is purely cost versus timing risk (see urgency and the risk-aversion parameter): trade fast and pay more impact, or trade slow and accept more exposure to random price drift. Add a signal — a belief the price will move in the trader's favor over the execution horizon — and a third force enters: alpha decay, the tendency for a short-term signal's predictive power to fade as other participants absorb the information. If the signal says "this stock is likely to rise over the next hour," every minute spent executing slowly also risks buying at a worse price as the anticipated rise happens before the order is fully filled — the delay isn't neutral risk anymore, it's actively working against the trade.

This reshapes the optimal schedule: rather than the smooth, symmetric curve a pure cost-minimization model produces, a signal-aware schedule typically front-loads execution, accepting higher impact to capture more of the position before the price moves as expected — the faster the signal decays, the more front-loaded the optimal schedule becomes.

Worked example: fast versus slow alpha

A trader has a signal suggesting a stock will rise about 20 basis points over the next 30 minutes, with the edge decaying roughly by half every ten minutes as other participants catch on. A cost-minimizing algorithm with no view on the signal might spread a 100,000-share buy evenly over 30 minutes, expecting 6 basis points of impact cost. A signal-aware algorithm instead front-loads: trading 50,000 shares in the first five minutes to capture more of the position before the anticipated rise, then trading the remainder more slowly as the signal's edge fades. If the rise materializes, the front-loaded schedule captures meaningfully more of the 20 basis-point move — even though its raw impact cost is higher, the combined effect of impact plus captured alpha comes out ahead.

What this means in practice

This is one of the main places execution and portfolio management overlap: a manager's conviction and time horizon on a signal should feed into the algorithm's urgency setting rather than being a separate decision, and many desks pass signal decay estimates directly into the scheduling logic. It also means execution quality can't be judged by impact cost alone on signal-driven trades — a schedule that "overpays" in impact but captures more of a real signal can still be the better outcome, which is why performance is usually measured against implementation shortfall (total cost including missed alpha) rather than impact cost alone.