PAX Gold liquidation map
Where leveraged PAXG positions would be force-closed. Longs liquidate below the current price, shorts above it, and each bar is the position value stacked at that level. This covers the accounts we can read, not the whole market, and the table under the chart says exactly how much that is.
Rules: a cluster is one bar (1% of price wide). “Nearest” is the largest bar within 15% of the mark. “Largest” is shown only when a bar holds at least 3% of its side. These describe where positions sit. They are not a forecast of where price goes.
How much of the market this is
Hyperliquid reports $23.8M of PAXG open interest per side. Positions can only be read account by account, and no list of all accounts exists, so this table reconciles what we plotted against that total. The last column is money we cannot see.
| Side | Plotted share of OI | Plotted | Beyond ±60% axis | No liquidation price | In accounts we cannot see |
|---|---|---|---|---|---|
| Longs | 13.5% | $3.2M | $1.8M | $4.4M | $14.3M |
| Shorts | 9.5% | $2.3M | $7.0M | $0 | $14.5M |
We draw a coin only when the accounts we can read hold at least 20% of open interest on each side. Known accounts skew toward larger traders, so the unseen remainder is not just a smaller copy of this chart. How accounts are found, and why coverage should grow →
Questions
What does the PAX Gold liquidation map show?
Each bar is the dollar value of open leveraged PAX Gold positions on Hyperliquid that would be force-closed if price reached that level. Long positions liquidate below the current price and shorts above it.
Is this every position?
No, and that is the most important thing on the page. Hyperliquid publishes positions per account, with no list of all accounts, so any map covers only the accounts its builder knows. We print our share of open interest beside the chart and withhold a coin when that share is too low to be meaningful.
Are these levels real orders?
No. They are liquidation prices of open positions, reported by the exchange itself. Nothing fires unless price actually arrives, and a level moves when a trader adds or removes margin.
Does price get pulled toward large clusters?
That is a popular claim and we have not seen it measured, so we make no claim about where price will go. We are now measuring it: every largest cluster we draw is recorded with a mirror band the same distance away in the opposite direction, and both are graded on the exchange’s own candles. If clusters do not get reached more often than their mirrors, we will publish that.
Do you only count recently-opened positions?
No. Every open position counts, however long ago it was opened. A liquidation price does not care about the age of the position behind it, and filtering by it would report a smaller number under the same name. The only age rule here is on our own reads: a position read more than 6 hours ago is dropped, because a stale read is a wrong price, not an old one.
How fresh is the data?
Reading every known account takes over an hour at the exchange's public rate limit, so this is a rolling snapshot, not a live feed. The age of the reads is printed above the chart, and positions read more than 6 hours ago are left out.
Why do the bar colours differ from other maps?
Ours are coloured by the actual leverage of the positions in each bar, as reported by the exchange. Some maps colour bars by the leverage that would be needed to liquidate at that price, which is a property of the price axis rather than of anyone's position.