Do longs or shorts get liquidated more? Hyperliquid data
Forced longs were 58.95% of Hyperliquid liquidated dollars over 114 days, but fell to 49.64% by September. See the months and what followed.
On Hyperliquid, longs were liquidated more in dollars, but not in every month. Across 176 smaller coins and 114 days (June 3 to September 24, 2026), forced longs made up 58.95% of liquidated dollars. By September the split was almost even, at 49.64% longs.
Over a longer span, from May 2025 to September 2026, 69.20% of flagged cascade bars were forced-long. This is a count of past cases, not a forecast.
What is a forced long, and what is a forced short?
A liquidation happens when an exchange closes a trade by force because the trader's money backing it ran out. A long is a trade that gains when price rises. A forced long is closed by selling. A short gains when price falls. A forced short is closed by buying.
A perp is a futures contract that never expires. All numbers here are for Hyperliquid perps. Our guide to how liquidation works explains when an exchange steps in.
How much more were longs liquidated?
It depends on how you count. We used three views of the same exchange, and each tilts toward longs.
| View | Forced-long share |
|---|---|
| Liquidated dollars, 176 coins, 114 days (June 3 to September 24, 2026) | 58.95% |
| Same, without ZEC | 68.20% |
| Days when longs beat shorts in dollars | 65 of 114 days |
| Flagged five-minute cascade bars, 197 coins (May 25, 2025 to September 30, 2026) | 69.20% of bars |
| Dollars in those flagged bars | 75.77% |
| The 300 largest liquidation episodes (June 2025 to September 2026) | 70% (210 of 300) |
The rest in each row were forced shorts.
The flagged-bar rows need care. A flagged bar is a five-minute stretch with unusually large forced closures on one side. One market-wide move can flag many coins and many bars, so these are shares of bars, not of separate crashes.
None of these sets include BTC, ETH, SOL or HYPE. They describe smaller coins.
ZEC matters. That one coin made up 35.86% of all liquidated dollars in the 114-day set. Without it, forced longs rose to 68.20% of dollars. ZEC had more forced-short dollars than forced-long dollars, so it pulled the total toward 50%.
Did the split change over time?
Yes, and by a lot.
| Month (2026) | Forced-long share of dollars | Without ZEC | Days longs beat shorts |
|---|---|---|---|
| June | 71.22% | 79.40% | 20 of 28 |
| July | 64.64% | 71.86% | 16 of 31 |
| August | 50.98% | 58.83% | 17 of 31 |
| September, to the 24th | 49.64% | 64.29% | 12 of 24 |
Four months is a small sample, so treat the slide as a description, not a trend. In the longer flagged-bar set, the forced-long share was above half in 16 of the 17 calendar months. The first month, May 2025, is partial.
| Flagged-bar month | Forced-long share |
|---|---|
| July 2025 | 70.84% |
| October 2025 | 77.98% |
| January 2026 | 70.40% |
| April 2026 | 53.58% |
| August 2026 | 46.43% |
| September 2026 | 55.99% |
August 2026 was the one month where shorts led. We did not test why the share moved. It could follow the mix of coins or what price did, and we cannot say which from these numbers.
Which side had the largest single minutes?
We ranked the biggest coin-minutes in the 114-day set. Of the top 20, 9 were forced longs and 11 were forced shorts. So the very biggest minutes were split about evenly, even though dollars overall tilted to longs.
The two largest were both ZEC.
- June 4, 2026, 02:03 UTC: $17,155,663 of forced longs in one minute.
- September 16, 2026, 18:08 UTC: $13,921,569 of forced shorts in one minute.
ZEC accounts for 13 of the top 20. The 20 largest coin-minutes made up 13.84% of all dollars, so a handful of minutes can swing a short study.
What happened after cascades led by each side?
This is a different set: the 300 largest liquidation episodes on Hyperliquid, from June 2025 to September 2026. We merged flagged bars within an hour of each other into one episode, and kept the largest bar. Our full write-up is in what happened after 300 liquidation cascades.
The median is the middle result after sorting. "Share up" is the share of cases where price was higher. All moves are before fees.
| After the biggest bar | Forced-long episodes (210): median, share up | Forced-short episodes (90): median, share up |
|---|---|---|
| 1 hour | +0.81%, 66.19% | −0.59%, 40.00% |
| 4 hours | +0.99%, 64.76% | −0.83%, 36.67% |
| 24 hours | +0.20%, 52.38% | −1.65%, 37.78% |
After forced long selling, the middle result was a rise. After forced short buying, it was a fall. A day later, the forced-long group was close to a coin flip at 52.38% up.
The cases disagreed a lot. For forced-long episodes at one hour, the 10th and 90th percentiles were −1.91% and +3.72%. A percentile is a cutoff: a tenth of cases fell below the lower number and a tenth rose above the upper one. For forced-short episodes they were −4.26% and +1.46%.
The side label is only the side of the single biggest bar. Of the 300 episodes, 196 had both longs and shorts forced out at different moments. And these episodes were chosen after the fact by size, so they are not a rule you could have followed live.
A fee check matters too. With an assumed 0.10 percentage point round-trip fee, an example long after the forced-long group had a one-hour median of +0.71%. An example short after the forced-short group had +0.49%. That ignores the gap between buy and sell prices, funding payments, and delay. It is not a return anyone earned.
What does this not show?
It does not show that longs are the more common trade. A liquidation count tells you how many trades were closed by force, not how many were open. If many more traders hold longs, they will show up more in liquidations.
It does not show how much leverage those traders used. For rough odds by leverage on Bitcoin, see how often 10x or 20x leverage would get liquidated.
It does not show a profit rule. It does not cover BTC, ETH, SOL or HYPE. And it cannot say why the long share fell during 2026.
How we measured this
For the dollar view, we added up Hyperliquid's own liquidation fills each minute for 176 coins, from June 3 to September 24, 2026 (114 days). Forced longs are the dollars sold by the exchange; forced shorts are the dollars bought. For the flagged-bar view, we marked five-minute bars where one side's liquidations were unusually large, on 197 coins from May 25, 2025 to September 30, 2026. For the 300 episodes, we required an exact price at every time we measured, filled in nothing, and left out any case with missing data.
Limits: the dollar set is 114 days and four months. Flagged bars cluster in time. Dollar totals depend heavily on a few huge minutes. The three sets use different coins and dates, so they should not be added together.
Try this in Comparand
With Comparand connected to your AI app, you can ask about Hyperliquid:
- "Show the largest liquidation minutes from June 1 to September 24, 2026, on the sell side." The ranking tool (hl_liquidations_top) returns the biggest single minutes in a span of up to 548 days, for all coins or up to 20 you list. You can limit it to the buy side or the sell side. It costs $0.01 per call on a pay-as-you-go key.
- "Show the cascade events on October 10, 2025, with what followed each one." The daily tool (hl_cascade_events with after=1) returns that archive day's cascade events and what the price did after each.
These tools return single minutes and events. Working out shares by side, or medians after each side, is a step you do yourself. You can try 3 coins free, read the data tool docs, or see pricing.
To see when these events cluster during the day, read what time of day liquidations happen.
The short version
- On Hyperliquid, forced longs were 58.95% of liquidated dollars over 114 days and 69.20% of flagged cascade bars from May 2025 to September 2026.
- The long share fell during 2026, from 71.22% in June to 49.64% in September, and shorts led the flagged bars in August 2026.
- The 20 biggest minutes were split 9 long and 11 short, and ZEC alone made up 13 of them.
- After the 300 largest episodes, the one-hour median was +0.81% after forced longs and −0.59% after forced shorts, before fees, with wide spread.
- Four months, a few huge minutes and no BTC, ETH, SOL or HYPE: past cases, not a forecast.
Questions people ask
Do longs or shorts get liquidated more in crypto?
On Hyperliquid, longs more often, but not every month. Forced longs were 58.95% of liquidated dollars from June 3 to September 24, 2026, and 69.20% of flagged cascade bars from May 2025 to September 2026. In September 2026 the dollar split was close to even, at 49.64% longs.
What does it mean when longs get liquidated?
A long gains when price rises. When a long runs out of money, the exchange closes it by selling. That label says which trades were closed. It does not say where price goes next.
Does price keep falling after long liquidations?
Not in our sample. In the 300 largest Hyperliquid liquidation episodes, the 210 led by forced longs had a middle move of +0.81% one hour later, before fees. About a third of them still fell. These are past cases, not a forecast.
Which was the largest liquidation minute we measured?
ZEC on June 4, 2026, at 02:03 UTC: $17,155,663 of forced longs in one minute. The second largest was ZEC forced shorts on September 16, 2026, at $13,921,569.