Drawdown Triggers and Review Thresholds

Every strategy, even a good one, will eventually go through a losing stretch — a drawdown — that has nothing to do with anything being broken. The hard part is deciding, in real time, whether a given drawdown is that kind of ordinary bad patch or an early sign of something genuinely wrong. Making that call in the moment, under the stress of watching a strategy lose money, is exactly when judgment is most likely to go wrong in either direction: cutting a strategy right before it recovers, or holding on to one that's quietly broken because "it's always come back before." Drawdown triggers exist to take that decision out of the emotionally charged moment and put it into a rule agreed on beforehand.

A typical setup defines several tiers. A first threshold — say, a drawdown reaching a level the strategy's own history says happens routinely — triggers nothing more than a documented review: has anything changed in the market environment, in the strategy's inputs, in execution costs? A second, deeper threshold triggers an automatic size cut, reducing the strategy's capital while the review continues, so that further losses at full size don't compound while the question is being investigated. A third, still deeper threshold — often set near the worst drawdown the backtest or live history has ever produced — triggers a full stop: the strategy is taken off risk entirely pending a decision from the capital allocation committee on whether, and how, it resumes.

Consider a strategy whose backtest shows a typical worst drawdown of around $150,000 and an absolute worst case of $400,000. The review threshold might be set at $150,000 (routine, expected, just look), the size-cut threshold at $250,000 (deeper than typical — cut size in half while reviewing), and the stop threshold at $400,000 (at the edge of anything the strategy's history has ever produced — halt and escalate). A strategy that draws down $180,000 triggers only the first-tier review; the same strategy at $420,000 triggers an automatic halt regardless of how convinced anyone currently is that "it'll come back."

What this means in practice

The value of setting these thresholds in advance is that they remove the single most common failure mode in managing a live strategy through a bad stretch: the tendency to keep raising your own tolerance in real time, waiting just a little longer each time the strategy loses a little more. A pre-agreed trigger doesn't require anyone to correctly judge, in the moment, whether this drawdown is different from the last one — the threshold was set using the strategy's actual history, before anyone had a stake in a particular answer.