Position Sizing, Drawdown Caps, and Strategy Promotion

Abstract

Position sizing can turn a promising strategy into an unsafe one. A candidate should not be promoted only because its best weight improves Sharpe. It also has to respect drawdown caps, concentration limits, and operational readiness.

For developers, weight selection should be a reproducible artifact.

The Frontier

A weight frontier tests the same candidate across a grid of allocations. Each point should report incremental PnL, book total PnL, daily and weekly Sharpe deltas, drawdown worsening, and robustness diagnostics.

The selected weight should be the best gate-passing weight, not simply the best headline metric.

Research Weight vs Operational Weight

Sometimes the mathematically best weight is not the right paper weight. A slightly lower rounded weight can preserve more drawdown cushion while keeping most of the benefit.

That distinction is healthy. Research can identify the frontier. Operations can choose a conservative value that is easier to monitor.

Promotion Requires More Than Return

A strategy should pass data integrity checks, executable contract checks, fill policy checks, robustness checks, and paper readiness checks. Return is one input. It is not the promotion policy.

Takeaway

Position sizing belongs in the promotion report. Developers should select weights under explicit caps and carry the chosen operational weight into the launch contract without changing hidden assumptions.