Up capture is the share of the market’s gain your portfolio earned during the months the market rose. Down capture is the share of the market’s loss it absorbed during the months the market fell. Both are expressed as percentages of the benchmark, where 100 means “exactly the market.”
Lemma Analysis computes them on monthly returns against SPY. Months are sorted by the sign of the benchmark, not of the portfolio; a month where SPY finished exactly flat belongs to neither group, because there is no market move to take a share of. Inside each group both sides are compounded, and the two compounded results are divided — not the month-by-month ratios averaged, which produces a number that matches nothing in the account.
A portfolio posts 104 / 78: in rising months it earned 104% of what the index earned, and in falling months it lost only 78% of what the index lost. That asymmetry — more of the upside, less of the downside — is the shape every active strategy is trying to buy.
The mirror image, 92 / 118, describes a portfolio that lags on the way up and falls harder on the way down. It can still show a positive year; it will still be a worse holding than the index.
A negative down capture is not a bug: it means the portfolio gained in months the market fell.
Read the pair, never one number alone:
| Shape | Reading |
|---|---|
| up > 100, down < 100 | genuine asymmetry — the profile worth paying for |
| up > 100, down > 100 | leveraged exposure: more market, not more skill — compare with beta |
| up < 100, down < 100 | defensive: less market. Fine if that is the mandate |
| up < 100, down > 100 | the worst of both — check concentration and costs |
Capture ratios say what beta says, but split into the half you wanted and the half you did not. A beta of 1.0 hides a 120 / 80 profile and an 80 / 120 profile equally well.
They matter most when judging a defensive allocation or a hedge. “It went up less” and “it went down less” are the same sentence about beta and two different sentences about capture — and only the second pair tells you whether the protection you paid for was worth its cost in rising markets.
Both numbers depend entirely on which months landed in which group, so a short history makes them unstable: five down months are not a description of how the portfolio behaves in a bear market. When the benchmark’s compounded return in a group is near zero, the ratio explodes — Lemma Analysis shows a dash instead of a number in that case.
They also say nothing about when the capture happened. A 78 down capture earned entirely in one quiet month tells you far less than the same figure spread over a real drawdown.