Maximum-drawdown budget

Definition

A maximum-drawdown budget declares, in advance, the largest peak-to-trough equity decline a strategy or account will accept — and binds sizing and de-risking rules to it. It converts the question “how much can we lose?” from a post-hoc discovery into a design constraint: every strategy’s size, the portfolio’s aggregation, and the de-risking schedule are derived from the budget, not vice versa.

How it works / structure

• Parameters (engine-executable): the budget (% from high-water mark), measurement basis (closed equity vs marked, daily marks pinned), the response curve — proportional de-risking as drawdown consumes budget (e.g. risk fraction scaled by remaining budget, the Grossman-Zhou shape) vs stepped triggers (at 50% of budget, halve size; at 100%, flatten and escalate to ratification), and the reset rule (new high-water mark vs time-based re-arm).
• Derivation chain: budget → tolerable per-trade f given streak arithmetic (risk-fixed-fractional) → per-strategy vol targets (risk-volatility-targeting) — the budget is the root node of the platform’s sizing tree.
• Daily companion: the session-scale version is a daily loss limit (same logic, one-session window — the load-bearing control of strategy-day-trading-styles); the budget is the campaign-scale control.
• Why pre-declared: decisions about cutting risk are worst when made inside the drawdown (bias-loss-aversion inverts discipline exactly then); the budget moves the decision to calm conditions.

When it applies

Every account and every ratified strategy on the platform carries one — the acceptance criterion is structural. Budgets also gate strategy comparison: replay drawdown vs budget is a pass/fail dimension independent of return.

Risk profile & failure modes

• De-risking lock-in: cutting size as drawdown deepens means recovering on smaller capital — proportional schemes mathematically slow recovery; the budget buys survival with recovery speed, a trade-off to acknowledge, not hide.
• Whipsaw at triggers: stepped de-risking at sharp boundaries flip-flops in choppy equity curves; hysteresis or proportional curves reduce it.
• Gap-through: like stops, budgets are intentions — a gap can consume several steps at once; aggregation honesty (risk-correlation-exposure) decides whether the budget was ever real.
• Budget theater: a budget without a wired response curve is a wish; the platform requires the response rules encoded, not implied.

Evidence & limits

Grossman-Zhou (1993) formalized drawdown-constrained investment and the proportional de-risking solution shape — the theory anchor. Drawdown statistics of specific strategies are replay outputs, not general facts. What is arithmetic: deeper drawdowns require disproportionately larger recoveries (a 50% loss needs +100%), which is the budget’s core justification.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Portfolio P under its 15% budget with proportional de-risking keeps replay max drawdown under 15% across this decade’s replay, including gap days” — falsified by the replay path.
• “The budgeted version of P retains at least 70% of the unbudgeted version’s replay CAGR” — falsified by the paired result.

Cross-references

• Derived controls: risk-fixed-fractional, risk-volatility-targeting, and per-session daily loss limits
• Sizing ceiling interaction: risk-kelly-criterion (fractional-Kelly drawdown relief)
• The failure it pre-empts: bias-loss-aversion (in-drawdown decisions)
• Aggregation honesty: risk-correlation-exposure, port-exposure-netting