Defining Sleeves in a Multi-Strategy Book
A sleeve is a bucket of capital and risk carved out for one coherent strategy inside a larger multi-strategy book, and drawing its boundaries correctly is what makes the rest of portfolio construction possible.
Prerequisites: The Multi-Strategy Platform Model
Before a firm can decide how much capital to give any one strategy, it has to answer a more basic question: where does one strategy end and another begin? A sleeve is the answer — a defined bucket of capital, risk, and positions assigned to one coherent strategy, tracked separately from every other sleeve in the book even though all of them ultimately share the same balance sheet.
Getting the sleeve boundaries right matters more than it sounds, because every later step — correlation estimates, risk budgeting, performance attribution — is computed per sleeve, and a badly drawn boundary corrupts all of them at once.
A sleeve is a bookkeeping and risk-management boundary, not a natural law of markets — where you draw it determines what correlations, drawdowns, and allocations look like downstream, so the boundary itself is a modeling choice worth getting right.
What makes a good sleeve boundary
A sleeve should group positions that share a genuine, common return driver and separate positions that don't, even if they trade the same instruments. A merger-arbitrage desk and a long/short equity desk that both happen to hold the same stock should still be two sleeves, because one's return depends on deal-completion risk and the other's depends on the stock's fundamentals — lumping them together would hide that their risks come from different places. Conversely, splitting one coherent momentum strategy into "momentum-tech" and "momentum-industrials" sleeves for no reason other than sector labels adds bookkeeping noise without capturing a real difference in what drives returns.
Worked example
A multi-strategy fund initially defines one "equities" sleeve covering all its stock positions, regardless of holding period or strategy. When performance dips, nobody can tell whether the problem is the fund's short-horizon mean-reversion book or its multi-month fundamental long/short book, because both are lumped into one number. The fund redefines its sleeves along the actual return driver: a "short-horizon mean-reversion" sleeve and a "fundamental long/short" sleeve. In the next drawdown, the redefined sleeves show the mean-reversion sleeve down 6% while the fundamental sleeve is flat — information the single combined "equities" sleeve had completely hidden, and information that directly tells risk management which book to cut.
What this means in practice
Sleeve definitions should be revisited periodically, not set once and left alone, because a strategy can evolve — a pod that started as pure statistical arbitrage might drift toward holding positions longer as its book grows, and at some point it stops behaving like the sleeve it was originally filed under. Firms that never re-examine sleeve boundaries end up making risk decisions based on categories that no longer describe what's actually happening inside them.
Defining sleeves by asset class or desk name rather than by return driver is the most common mistake — two strategies can trade the exact same instruments for entirely different reasons, and grouping them together because they "are both equities" hides the diversification (or lack of it) that actually matters for risk.
Further reading
- Ineichen, Asymmetric Returns (ch. 4, on strategy classification)