One question, one answer
Are longs or shorts getting liquidated more right now?
The side that is losing, measured rather than guessed, from nine exchanges' own liquidation feeds.
Longs, 83% of the damage
Of the $628.6 million liquidated across 10 exchanges in the last 24 hours, $521.8 million (83%) was long positions and $106.8 million (17%) was shorts. Leveraged buyers were the crowded side and price came down into them.
| Exchange | 24h liquidated | Longs | Shorts |
|---|---|---|---|
| Binance | $273.8 million | 86.5% long | 13.5% short |
| Gate | $125.5 million | 96.1% long | 3.9000000000000057% short |
| Hyperliquid | $86.0 million | 57% long | 43% short |
| Bybit | $63.1 million | 85.3% long | 14.700000000000003% short |
| OKX | $53.5 million | 78.8% long | 21.200000000000003% short |
| HTX | $12.3 million | 83.9% long | 16.099999999999994% short |
| Binance (coin-M) | $11.2 million | 96.5% long | 3.5% short |
| Bitfinex | $399K | 99.2% long | 0.7999999999999972% short |
The live liquidation feedBy exchangeLong/short account ratiosLiquidation heatmap
What the split actually tells you
It tells you which way the crowd was leaning when the move came, and that is usually more useful than the size of the move itself. Heavy long liquidations mean leveraged buyers were crowded and price went down into them. Heavy short liquidations mean the squeeze ran the other way. When the two sides are close to even, the market chopped rather than trended, and both sides paid for it.
The one thing the split does not tell you is what happens next. A wave of long liquidations often marks the end of a flush, because the leveraged supply has already been taken out - and just as often it is the first leg of a longer one. Read it as a description of what has already happened to positioning, not as a signal.
Why it differs so much between exchanges
The per-venue table above usually disagrees with itself, and that is the interesting part. A venue whose 24-hour flow is ninety percent long liquidations while another sits near fifty is telling you its own users were positioned differently - often a retail-heavy venue against one with more professional flow, or a regional user base that was long into news the rest of the market was not. The market-wide figure hides that; the venue column is where it shows.
Method, in full
MarginPad runs its own collector subscribed to the public liquidation websocket of nine exchanges: Binance (USD- and coin-margined), Bybit, OKX, Hyperliquid, Gate, HTX, dYdX, BitMEX and Bitfinex. Every forced close is normalised to symbol, side, fill price and dollar notional, then aggregated over a rolling 24-hour window. These are observed events, not a model and not a vendor feed. Nothing is filled in when a venue goes quiet, and if the collector itself is down the page says so rather than showing yesterday's figures as if they were current.
Take the data
Free as JSON, no key: GET https://marginpad.io/api/v1/liquidations carries data.market.long and data.market.short plus the same split per coin, and GET https://marginpad.io/api/v1/venues carries it per exchange with a longPct field. Both stamp the measurement time in ts.
Questions
Are longs or shorts liquidated more often?
Over a long enough window longs, by a wide margin, for a simple structural reason: crypto spends more of its time trending up than down, so more leverage sits on the long side and is therefore available to be liquidated. Over any single day it goes either way, which is why the figure at the top of this page is measured for the trailing 24 hours rather than stated as a rule.
What does it mean when long liquidations spike?
That leveraged buyers were crowded and price fell into their liquidation levels, forcing market sells that push price further down - the cascade. The useful follow-up question is whether open interest fell with it. Price down with open interest down means leverage was flushed out; price down with open interest flat means new shorts replaced the liquidated longs.
Does a short squeeze show up here?
Yes, as the mirror image: a run of short liquidations, concentrated in a short window, usually on the venues where shorts were most crowded. The live feed at /rekt/ shows the sequence as it happens, which is the clearest way to tell a squeeze from steady two-way flow.
Where does this data come from?
MarginPad's own collector, subscribed directly to the public liquidation websocket of nine exchanges. These are observed forced closes rather than estimates, and the venue set is published so the number can be checked against its own source.
Can I get it as an API?
Yes, free and keyless: /api/v1/liquidations for the market-wide and per-coin split, /api/v1/venues for the same split per exchange. Both are documented on the free crypto API page.
Other questions we answer with a live number
Related: the live liquidation feed · by exchange · long/short account ratios · liquidation heatmap