The comparison only exists when the evidence does.

Return of the entity minus the return of holding BTC over the same window, net of entity fees. Positive means the entity beat the benchmark.A market-neutral strategy losing to BTC in a bull run is not underperforming. Meaningless without beta: a 3x leveraged long on the benchmark shows enormous positive alpha in a rising market and enormous negative alpha in a falling one, and neither is manager skill. Read beta_* and r_squared_* alongside this, and strategy_category, before treating it as a verdict. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Return of the entity minus the return of holding ETH over the same window, net of entity fees. Positive means the entity beat the benchmark.Meaningless without beta: a 3x leveraged long on the benchmark shows enormous positive alpha in a rising market and enormous negative alpha in a falling one, and neither is manager skill. Read beta_* and r_squared_* alongside this, and strategy_category, before treating it as a verdict. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Return of the entity minus the return of holding SOL over the same window, net of entity fees. Positive means the entity beat the benchmark.Meaningless without beta: a 3x leveraged long on the benchmark shows enormous positive alpha in a rising market and enormous negative alpha in a falling one, and neither is manager skill. Read beta_* and r_squared_* alongside this, and strategy_category, before treating it as a verdict. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Return of buying BTC on the window start date and holding to the window end date, less one entry swap cost of 10 basis points.Same start and end dates as the entity window, so the two are directly comparable.

Return of buying ETH on the window start date and holding to the window end date, less one entry swap cost of 10 basis points.Same start and end dates as the entity window, so the two are directly comparable.

Return of buying SOL on the window start date and holding to the window end date, less one entry swap cost of 10 basis points.Same start and end dates as the entity window, so the two are directly comparable.

Slope of the entity's periodic returns regressed on BTC's over the same window and the same intervals. 1 tracks the benchmark, 3 is roughly three times geared, 0 is market-neutral, negative is short.Read with r_squared_btc: a beta of 3 explaining 98% of variance is a leveraged tracker, while the same beta explaining 5% is two noisy series coinciding. Undefined (null) with fewer than three paired return intervals, or when BTC did not move over the window — null means "cannot say", not zero. On a downsampled series the figure is an estimate: a product rebalanced to 2x daily reads about 2.11 when measured on alternate days, because two geared daily steps do not compose into one geared two-day step. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Slope of the entity's periodic returns regressed on ETH's over the same window and the same intervals. 1 tracks the benchmark, 3 is roughly three times geared, 0 is market-neutral, negative is short.Read with r_squared_eth: a beta of 3 explaining 98% of variance is a leveraged tracker, while the same beta explaining 5% is two noisy series coinciding. Undefined (null) with fewer than three paired return intervals, or when ETH did not move over the window — null means "cannot say", not zero. On a downsampled series the figure is an estimate: a product rebalanced to 2x daily reads about 2.11 when measured on alternate days, because two geared daily steps do not compose into one geared two-day step. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Slope of the entity's periodic returns regressed on SOL's over the same window and the same intervals. 1 tracks the benchmark, 3 is roughly three times geared, 0 is market-neutral, negative is short.Read with r_squared_sol: a beta of 3 explaining 98% of variance is a leveraged tracker, while the same beta explaining 5% is two noisy series coinciding. Undefined (null) with fewer than three paired return intervals, or when SOL did not move over the window — null means "cannot say", not zero. On a downsampled series the figure is an estimate: a product rebalanced to 2x daily reads about 2.11 when measured on alternate days, because two geared daily steps do not compose into one geared two-day step. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Observed span of the window in days, from the first to the last available observation inclusive.Lower than window_days means the entity has less history than the window asked for.

Whether a fee haircut was subtracted from the reported twr. True when the venue reports gross performance and a fee schedule was on file for the entity.False has two very different meanings: the venue already reports net of fees, or no fee schedule is recorded and the figure is gross. Read it with nav_quality before comparing two entities from different venues.

Headline entity return minus the median realised depositor return. Positive means the advertised number flattered the vault relative to its own investors' outcomes.Requires a depositor cross-section, so it exists only for venues that publish one (Hyperliquid followers, OKX copy traders). Depositor returns are since their own entry, not over this window, so treat the gap as directional rather than exact.

Median of all_time_pnl / (equity - all_time_pnl) across the depositor cross-section observed on the window end date. What the typical investor actually realised.Only depositors with a positive implied cost basis are counted; fully withdrawn depositors cannot be computed and are excluded. The cross-section decays as depositors exit, so it is only ever as complete as the day it was captured.

Whether this figure may be ranked against other entities. False for nav_quality='roi', and false for venues whose input semantics are not yet verified.An ineligible row is still a real measurement of that entity — it is simply not comparable to the others. Show it on its own card, never in a league table.

True when the entity has history covering the whole requested window. False means the figures are since inception and must be labelled as such.Never compare a false row against a true row of the same window_days.

Largest peak-to-trough decline of the per-unit value series within the window, as a non-negative fraction of the peak.A drawdown computed from a downsampled series understates the true figure, possibly badly, because the troughs between observations are not in the data. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

'reported' when the venue published a true per-unit NAV or share price; 'derived' when the per-unit series was reconstructed from account value net of flows; 'roi' when the venue published only a money-weighted return.'roi' rows are excluded from headline rankings. Mixing money-weighted ROI with time-weighted return is the error this column exists to prevent. A venue-published share price is more trustworthy than our reconstruction, not less.

Share of the entity's return variance explained by BTC over the window, from 0 to 1. This is what tells you whether the beta is gearing or coincidence.A high value does not mean good or bad, only that the benchmark explains the movement. Null exactly when beta is null. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Share of the entity's return variance explained by ETH over the window, from 0 to 1. This is what tells you whether the beta is gearing or coincidence.A high value does not mean good or bad, only that the benchmark explains the movement. Null exactly when beta is null. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Share of the entity's return variance explained by SOL over the window, from 0 to 1. This is what tells you whether the beta is gearing or coincidence.A high value does not mean good or bad, only that the benchmark explains the movement. Null exactly when beta is null. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

'daily' when every observation in the window came from a same-day read; 'downsampled' when any observation came from Hyperliquid's coarse allTime series.A mixed window degrades to downsampled — it is only as good as its coarsest point.

Return of the entity over the window, computed from its per-unit value series as v[end]/v[start] - 1 and therefore unaffected by deposits and withdrawals. Net of the venue fees recorded for the entity.Comparable across entities when nav_quality is 'reported' or 'derived'. Rows with nav_quality='roi' are venue-published money-weighted returns, are not time-weighted, and are excluded from headline rankings. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.

Sample standard deviation of per-step returns within the window, scaled to a year using the observed mean spacing between observations rather than an assumed daily cadence.Scaling an irregular ~biweekly series is approximate. Unreliable when sampling='downsampled': backfilled Hyperliquid history is ~93 points across a vault lifetime, so intermediate moves are absent.