Leverage over one week

Hold for a week and the top half of the list is no longer available to you. The best multiplier still standing over one week is 5x, on 24 of 45 markets, and 34 of 45 markets sat at or below it.

Every market, weekly multiplier and weekly move

Long side and short side are measured separately and the worse of the two sets your multiplier, which is what makes one figure valid whichever way you go. The gap between them is the market's weekly asymmetry: S&P 500 ETF shows 4.16 per cent against 3.81 per cent, while Hyperliquid (HYPE) shows 21.7 per cent against 26.8 per cent.

S&P 500 ETF carried the highest weekly multiplier at 20x, holding 5.00 per cent of room against 4.16 per cent of movement. Hyperliquid (HYPE) carried the lowest at 2x, with 50.0 per cent against 26.8 per cent. Across the whole set the spread runs 2x on 10, 5x on 24, 10x on 10, 20x on 1.

Carry is left out of every weekly figure

Hold for a week on the flat-fee venue and you cross 21 rolling-fee intervals at a rate that venue does not publish, so no figure here includes it and every cost here is a floor. The notional-fee venue publishes funding hourly, capped at 4.00 per cent per hour with an interest component of 0.00125 per cent per hour: funding documentation. Carry is therefore not comparable across the two venues, because a comparison needs both rates and 1 of the 2 is unpublished.

Your other weekly cost is the maintenance rule. The notional-fee venue holds half the initial margin at the asset ceiling, which removes half your room there: liquidation documentation, with per-asset ceilings from the venue's own asset list. The flat-fee venue holds no maintenance buffer and caps your loss at the wager, so your room is one part in the multiplier and your worst case is the wager itself.

Weekly figures come from daily bars measured against Moon's published multipliers, which is a different claim from Moon listing the market: all 50 tracked markets are confirmed at instrument level. Daily bars from Twelve Data, venue terms from registry 2026-08-28a.

Method, restated because the weekly numbers are the ones most often quoted without it. A multiplier survives the week when 95 per cent of the sampled five-day windows would not have liquidated at it, on the worse of the two sides.

The weekly threshold is therefore the 95th percentile of about 396 five-day windows per market, and the multiplier is the last one on the venue's published list whose room clears it. That leaves roughly 1 week in 20 beyond the threshold. Registry 2026-08-28a, reviewed through 2026-11-28.

Window counts are not uniform: the typical market contributes 396 weekly windows and Gram contributes 12, so its multiplier rests on the shortest sample in the table. Shorter holds: one hour to one session, one minute to five minutes. Procedure: how to choose. Stamped 2026-08-28 06:26 UTC, 45 of 45 markets with enough daily bars for a weekly window. Bars from Twelve Data.

Figures on this page were rebuilt 2026-08-28 by the MarketMoves editorial desk, from the sources named in the methodology. Nothing here is a forecast, and none of it is advice.