Reallocating Based on Recent Performance

A sleeve that just had its best quarter in two years is an obvious candidate for more capital. A sleeve that just had its worst quarter is an obvious candidate for less. Both instincts feel like ordinary risk discipline, and sometimes they're correct — a real edge that just showed up should be sized up, and a real problem that just showed up should be sized down. But recent performance is a mix of skill and noise, and reallocating purely on the recent number, without asking which one dominates, tends to add capital right as a lucky streak is about to mean-revert and cut capital right as a rough patch is about to turn around.

Reallocating based on recent performance describes this common practice and the specific trap hidden in it: strategy-level performance chasing, which behaves exactly like the well-documented mistake of chasing a hot mutual fund, just one level up in the hierarchy.

Why the pattern is self-defeating

If sleeve returns have any tendency to mean-revert over the horizon being used to reallocate — which many strategies genuinely do, since a strategy's edge doesn't usually change dramatically quarter to quarter — then a rule that adds capital after a strong quarter and removes it after a weak one is, on average, adding capital right before reversion pulls performance back down and removing capital right before reversion pulls it back up. The rule looks disciplined and feels responsive, but it's mechanically working against the sleeve's own statistical tendencies.

Worked example

A firm's rules-based process cuts a sleeve's allocation by 25% whenever it has a losing quarter and adds 25% whenever it has a winning one. A statistical-arbitrage sleeve has a rough quarter driven by an unusually crowded unwind across the industry — a known, largely one-off event rather than a change in the sleeve's underlying edge — and the firm cuts its allocation by 25% right after. Over the following two quarters, the same sleeve recovers strongly as the crowding unwinds, but the firm is running it at 75% of its prior size and captures a correspondingly smaller share of the recovery. Had the firm instead investigated why the quarter was bad before resizing, it likely would have distinguished a temporary, market-wide event from a genuine breakdown in the strategy.

What this means in practice

The fix is not to ignore recent performance — a real change in edge is real information — but to ask what's driving it before reallocating: is there a specific, identifiable reason the strategy's edge changed (a crowded trade unwinding, a data source degrading, a market regime shift the strategy wasn't built for), or is the move within the normal range of noise for a strategy of that volatility and time horizon? Reallocation decisions built on that diagnosis, rather than on the raw number alone, avoid mechanically selling low and buying high at the sleeve level.