# JUP liquidation heatmap · 24H

Estimated leverage parked at every price, hour by hour, with the liquidations that actually fired drawn on top. Recorded by our own collection network across the major perpetual venues.

Source: https://bykaranteli.com/liqmap/jup · attribute to ByKaranteli and quote the timestamp.
Snapshot taken 2026-10-17T14:53:01.411Z.

Coverage: Venues contributing: 11 of 11 · all fresh · 6 model venues · real prints from 11 recorded liquidation feeds, 57% of the estimated 24h total is recorded · as of 14:52 UTC.

$12K of JUP shorts liquidate at 0.3054 (2.7% above); $37K of longs at 0.2958 (0.5% below).

Window: 24H (public view). Price at the snapshot: 0.2974 USD (JUPUSDT).

## Nearest clusters

Estimated notional that would be forced out at each level.

| Cluster | Price (USD) | Distance from price (%) | Estimated size (USD) |
| --- | ---: | ---: | ---: |
| Cluster above (shorts) | 0.3054 | 2.70 | 12097 |
| Cluster below (longs) | 0.2958 | 0.53 | 36818 |

## Fuel and recorded liquidations

| Long fuel within 3% below (USD) | Short fuel within 3% above (USD) | Fuel skew | Liquidated 24h longs (USD) | Liquidated 24h shorts (USD) |
| ---: | ---: | --- | ---: | ---: |
| 706386 | 125012 | Longs 5.7× more | 23146 | 946 |

Fuel skew: estimated long fuel below vs short fuel above, within 3% of price. Liquidated totals: recorded from exchange streams, not modelled.

## Recent sweeps

Clusters the price actually moved through, read from the hourly archive.

| At (UTC) | Pocket | Price (USD) | Size (USD) | Now |
| --- | --- | ---: | ---: | --- |
| 2026-10-17 13:00 | long pocket | 0.2973 | 51948 | tested and rejected |
| 2026-10-17 12:00 | long pocket | 0.2999 | 68168 | still beyond it |
| 2026-10-16 21:00 | short pocket | 0.3017 | 32752 | tested and rejected |
| 2026-10-16 17:00 | long pocket | 0.2992 | 55835 | still beyond it |

24 hourly frames so far; the archive deepens every hour.

How to read this: The heatmap is an estimate of where leveraged positions would be forced out, built from recorded taker flow and open interest across the major perpetual venues; levels the price has already swept are removed. It is not an exchange liquidation order book. The orange ticks and the 24h totals are real liquidations recorded from exchange streams. Each level is spread over a band that widens with its distance from price, because leverage is a distribution rather than five round numbers; the 30 day view averages the archive into four hour columns. The public view is the 24 hour window; members get the deeper windows, filters and playback.

## How to read LiqMap: four patterns, measured

Four shapes people look for on a liquidation map, each measured on our own hourly archive: how far the price moved 1, 4 and 24 hours after the shape appeared, how often it kept going the way the shape points, and the same numbers over every hour as the base rate. It records what the price did next, not why, and it is no forecast.

- **Cluster within reach**: The largest cluster on either side of the map sits within 1.5% of price. The pattern points toward that cluster. Compare the median move with the any hour figure to see whether the price moved more than usual after a large cluster came this close.
- **Lopsided fuel**: Estimated long fuel within 3% below price and short fuel within 3% above differ by a factor of 1.5 or more. The pattern points toward the heavier side. Continued next to the base rate shows how often the price moved toward the heavier side, against how often any hour rose.
- **Thin side**: One side of the map holds no cluster within 3% of price. The pattern points toward that empty side. The median move shows how far the price travelled after one side of the map was empty near price, against every hour without the filter.
- **Swept level**: The price crossed a level that the previous hourly frame marked as a cluster. The pattern points the way the price crossed. Reverted in 4h shows how often the price was back across the swept level four hours later.

### Price moves after each pattern

| Pattern | Events | Median move 1h | Median move 4h | Median move 24h | Base rate | Continued | Reverted in 4h |
| --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
| Cluster within reach | 263 | 0.57% (any hour 0.61%) | 1.25% (any hour 1.18%) | 3.45% (any hour 3.55%) | 50.0% | 47.0% (median -0.09%) | · |
| Lopsided fuel | 585 | 0.67% (any hour 0.61%) | 1.25% (any hour 1.18%) | 3.58% (any hour 3.55%) | 50.0% | 46.4% (median -0.14%) | · |
| Thin side | 112 | 0.78% (any hour 0.61%) | 1.43% (any hour 1.18%) | 3.45% (any hour 3.55%) | 50.0% | 60.4% (median +0.43%) | · |
| Swept level | 273 | 0.82% (any hour 0.61%) | 1.56% (any hour 1.18%) | 4.10% (any hour 3.55%) | 50.0% | 47.1% (median -0.14%) | 42.2% |

- Median move: the median size of the price change 1, 4 and 24 hours after the pattern, either direction; below it, the same median over every hour of the same coins.
- Continued: the share of events in which the price kept going the way the pattern points over 4 hours; below it, the median 4 hour move in that direction (+ is the pattern's way).
- Base rate: the share of all hours after which the price rose over the next 4 hours, the coin flip to hold Continued against.
- Reverted in 4h: swept levels only, the share of events with the price back across the level 4 hours later.
- A pattern with fewer than 30 events shows no number; · marks a value without data.

152,511 map frames of 201 coins since 2026-08-06; computed 2026-10-17 14:41 UTC. A frame is written each hour and reads the price about 40 minutes into the hour; a move is the change in the perpetual's last price from one frame to a later one.

BTC, ETH and SOL have hourly frames since 2026-08-06; every other coin since 2026-09-20, with one frame every six hours before that.

## Related

- https://bykaranteli.com/liqmap/btc
- https://bykaranteli.com/liqmap/algo
- https://bykaranteli.com/liqmap/sand
- https://bykaranteli.com/liqmap/hbar
- https://bykaranteli.com/liqmap/trump
- https://bykaranteli.com/liqmap/bome
- https://bykaranteli.com/liqmap/ethfi
- https://bykaranteli.com/liqmap
- https://bykaranteli.com/methodology

Everything in one fetch: https://bykaranteli.com/llms-full.txt