What does a survivable multiplier of 5x over one week actually mean?

That 95 per cent of the 396 or so weekly windows measured per market never moved far enough against the entry to liquidate you at it. It describes the sample of bars held here rather than forecasting the next week, and one window in 20 in that same sample did liquidate.

Why does the safe multiplier fall as you hold for longer?

Your distance to liquidation is fixed by the multiplier while the move against you grows with time. The median market's worst move in 1 window in 20 is 0.17 per cent at one minute and 18.2 per cent at one week, so the multiplier has to come down 5 steps across that range to keep covering the same share of windows.

Can you compute a multiplier for a hold under one minute?

No. The data vendor's finest granularity is a one-minute bar, so a sub-minute holding period cannot be measured with this dataset even though the platform's chart offers shorter intervals. That gap is a limit of the input, and it is reported rather than interpolated.

Does the cheapest multiplier to open also survive longest?

The two are unrelated on a flat-fee platform, where a $100 wager opens for $1.00 at 2x and $1.00 at 1000x. On a notional-fee platform your cost rises with the multiplier, and the two bases cross at 23.1x taker and 69.4x maker.

How much leverage your holding period supports

Decide how long you will hold, then pick the multiplier. Hold for one minute and 200x came through on 21 of 45 markets. Stretch the same position to one week and the best multiplier still standing is 5x, 5 steps down the 9 that platform offers.

The shorter list moves the same way: 10x over one minute on 11 of 17 listed markets, 3x over one week on 13 of 16. Do not carry a multiplier from one platform across to the other, because it caps leverage per asset between 2x and 40x and holds a maintenance buffer the 1000x platform does not.

5 multipliers go between a minute and a week

Holding period	Moon multiplier	Moon markets agreeing	Hyperliquid multiplier	Hyperliquid markets agreeing	Moon range
one minute	200x	21 of 45	10x	11 of 17	200x on 21, 500x on 20, 1000x on 4
five minutes	200x	20 of 45	10x	11 of 17	50x on 1, 100x on 20, 200x on 20, 500x on 3, 1000x on 1
one hour	20x	20 of 45	10x	11 of 17	10x on 3, 20x on 20, 50x on 16, 100x on 4, 200x on 2
one session	20x	21 of 45	10x	11 of 17	5x on 2, 10x on 11, 20x on 21, 50x on 8, 100x on 3
one week	5x	23 of 44	3x	13 of 16	2x on 10, 5x on 23, 10x on 10, 20x on 1

That column is a median and the spread behind it is wide. Hold one market for one length of time and take the multiplier from that market's own row: at one hour the highest was 200x and the lowest 10x.

Moon publishes asset classes rather than instrument lists, so every Moon figure above is a multiplier Moon offers measured against a market's own bars. All 45 tracked markets are confirmed at instrument level. Bars from Twelve Data.

Moves grow faster than the multipliers can absorb

Holding period	Median move against you	Worst in 1 in 20	Worst in 1 in 100	Median windows	Markets
one minute	0.04%	0.17%	0.35%	900	45
five minutes	0.10%	0.46%	0.92%	896	45
one hour	0.43%	2.09%	2.60%	841	45
one session	2.10%	3.01%	3.34%	511	45
one week	4.68%	18.2%	26.4%	396	44

Each row is a median, so it describes a typical market rather than the worst one. The move that hits you in the worst 1 window in 20 grows from 0.17 per cent over one minute to 18.2 per cent over one week, and the platform has only 9 multipliers to give back, so size for that bad window rather than for the median.

How these figures were built

A multiplier survives a holding period when 95 per cent of the sampled windows of that length would not have liquidated at it, on the worse of the long side and the short side.

Every answer is a multiplier a platform sells: 9 from 2x to 1000x on the flat-fee platform, 7 from 2x to 40x on the notional-fee platform. Room comes from the platform registry. The flat-fee platform holds no maintenance buffer, so room is one part in the multiplier, and the other holds half the initial margin.

Windows per market: 900 at one minute, 511 at one session, 396 at one week, so the 95th percentile leaves about 1 window in 20 past the threshold. Across the 3.29 decades from one minute to one week the central multiplier falls 1.52 steps per tenfold increase. Registry 2026-08-28a, reviewed through 2026-11-28.

Pick the page that matches your hold

Scalping

One minute to five minutes. Most markets took 200x, plus the ten extremes.

Day trading

One hour to one session. Every market at both horizons, 20x in the middle.

Swing trading

One week, from daily bars. 5x across 44 markets, the longest hold measured.

How to choose

Four steps, ending in what your multiplier costs on each platform and the 23.1x crossover.

Opening cost is the taker cost on a $100 wager and leaves out carry, so every cost here is a floor. The flat-fee platform charges every 8 hours at a rate it does not publish, and the other caps hourly funding at 4.00 per cent: funding documentation.

Tables stamped 2026-09-16 07:24 UTC. 45 markets have pages in this build and 45 have enough bars for a one minute multiplier.

Figures on this page were rebuilt 2026-09-16 by the MarketMoves editorial desk, from the sources named in the methodology. Nothing here is a forecast, and none of it is advice.