Reallocating Capital Between Strategies
How a desk decides to shift capital from a strategy that's underperforming toward one that's working — and why doing it too fast or too slow are both common, expensive mistakes.
Prerequisites: Every Trade Competes For The Same Capital
A desk running multiple strategies is constantly deciding, implicitly or explicitly, how much capital each one gets, and that allocation isn't static — strategies go through periods of outperformance and underperformance, and the natural question is whether to shift capital toward the strategy that's currently working and away from the one that isn't. Done well, this is one of the more valuable things a desk can do; done reflexively, it's a good way to consistently sell what's about to work and buy what's about to stop.
The timing problem at the center of reallocation
The central difficulty is that recent performance is a noisy signal for future performance, and the noisiest part of any strategy's return history is usually the most recent stretch. A strategy that's had a strong month might be demonstrating a genuine edge reasserting itself, or might simply be at the favorable end of normal variance that will revert. Moving capital based purely on trailing returns, without a view on why performance changed, tends to chase whichever strategy most recently got lucky and starve whichever one is due for a bounce — the opposite of what reallocation is supposed to achieve.
The more defensible version of reallocation ties the decision to something other than the raw P&L number: has the market regime shifted in a way that structurally favors one strategy over another, has a strategy's edge genuinely decayed (more competition, a changed market structure, a broken assumption), or has one strategy simply been running larger and less-diversified risk than its allocation implies. A desk that reallocates on these grounds is responding to a changed environment; a desk that reallocates purely on trailing Sharpe ratio over the last month is, more often than not, just performance-chasing with extra steps.
A concrete example: two strategies each run $50m. Strategy A is up 8% over the past quarter, Strategy B is flat. If the desk can point to a specific reason — Strategy B's edge relies on a spread that's structurally compressed and unlikely to widen again soon — reallocating $15m from B to A is a reasoned decision. If the only evidence is the quarter's P&L difference itself, with no view on why it happened, the same reallocation is a bet that recent performance predicts future performance, which for most strategies over a single quarter it does not.
Reallocating capital toward a recently strong strategy and away from a recently weak one is only sound if there's a reason beyond the trailing P&L itself — a regime shift, a decayed edge, or a risk mismatch. Reallocating on trailing performance alone tends to buy what just got lucky and sell what's due to mean-revert.
Further reading
- Grinold and Kahn, Active Portfolio Management