Concentration Caps On A Live Book

The trades a trader is most tempted to oversize are exactly the ones they are most confident in — which sounds like it should be rewarded, and mostly should be, until you remember that being wrong about a high-conviction idea is indistinguishable, from the position's point of view, from being wrong about a low-conviction one. A concentration cap is the desk's way of saying: no matter how sure you are, no single name gets to be large enough to sink the book by itself.

What the cap actually protects against

It is not protecting against ordinary volatility — the risk-budget ceiling from your stop distance already handles that. A concentration cap protects against the tail: a fraud, an accounting restatement, an acquisition that falls through, a regulatory action — events where the stock gaps 40–60% overnight and your stop does nothing, because there was no liquidity to exit into on the way down. A position sized at 3% of book on a normal-volatility basis can still be an existential position on a gap-risk basis, and the cap is set with that gap in mind, not the day-to-day standard deviation.

Typical caps run 3–5% of NAV per single name for a diversified long-short book, tighter for less liquid names and sometimes wider for the very largest, most liquid mega-caps where a 50% overnight gap is close to unheard of.

Worked example

A $120m book runs a 4% single-name cap, so $4.8m maximum in any one name. Your highest-conviction idea of the year — every thesis leg confirmed, stop is tight, liquidity is enormous — computes to a risk-budget ceiling of $7.1m using the ordinary sizing arithmetic.

The concentration cap binds first: you are capped at $4.8m regardless of how good the risk-budget math looks, because the risk-budget math is answering "how much can I lose to a normal move" and the concentration cap is answering "how much can I lose if this name gaps on news no stop could have caught."

A trader who overrides the cap because "this is my best idea" is not taking more risk on a good trade — they are taking on tail risk that the position-sizing framework was never designed to price, since normal-volatility sizing has nothing to say about a name-specific gap event.

Concentration is also a portfolio-level question

A single position under the per-name cap can still push the book into effective concentration if several capped positions are all exposed to the same underlying risk — three different "cheap regional bank" longs each at 4% add up to 12% of correlated exposure even though each individually satisfies the rule. The per-name cap is necessary but not sufficient; sizing correlated positions together is the check that catches what the per-name cap alone misses.