Transparency is a feature. This page explains what up/down capture measures and why — in plain terms. The precise formulas and numeric conventions are documented in our internal methodology; the summary here is deliberate, not an omission.
Capture describes how much of a benchmark's move an asset "captured" — separately for the periods when the benchmark rose and the periods when it fell. It is a way of scoring offense and defense against a benchmark rather than in absolute terms. The asset and benchmark series are aligned strictly period by period and must cover the same periods; a mismatch is rejected rather than guessed at. Inputs are simple period returns expressed as decimal fractions.
We use the arithmetic mean-ratio convention throughout: split the aligned history by the sign of the benchmark, then compare the asset's average move to the benchmark's average move within each group. A geometric (compound) alternative exists and is a common choice elsewhere, but we deliberately keep one convention end to end and state it plainly. Periods where the benchmark is exactly flat count as neither up nor down and are left out of both groups.
Up capture compares the asset's average return to the benchmark's average return across only the periods when the benchmark was up. A value above one means the asset, on average, out-gained the benchmark in rising markets; below one means it lagged. It is signed — if the asset was actually down on average while the benchmark rose, the figure comes out negative.
Down capture does the same thing across only the periods when the benchmark fell. Because the benchmark's average is negative in those periods, an asset that falls less than the benchmark scores between zero and one (good defense), while one that falls more scores above one.
The capture ratio folds offense and defense into a single figure by comparing up capture to down capture. Capturing more of the upside and less of the downside both push it higher, so a larger ratio is generally the more attractive profile.
