You are not liquidated at the price on your chart

Cost to Hold · No. 0004 · Friday 14 August 2026 · four minutes

The number that closes your position is not the number on your screen, and on a bad day it is not even a number anyone traded at. Today: what actually triggers a liquidation, the mechanism that can close you out because you were right, and a tape where bitcoin funding ran from −1.73% to +10.95% across four venues while open interest climbed on every one of them.

You are not liquidated at the price on your chart

The price moving on your screen is the last traded price — whatever the most recent trade on that venue happened to print. It is not what closes you.

Liquidation triggers on the mark price, which is a different number, computed differently, and deliberately not the same thing.

Hyperliquid documents its own construction precisely, so use theirs. Mark price is the median of three inputs: an oracle price plus a 150-second exponential moving average of the gap between Hyperliquid's mid and that oracle; the median of Hyperliquid's own best bid, best ask and last trade; and the median of the perp mid prices on Binance, OKX, Bybit, Gate.IO and MEXC, weighted 3, 2, 2, 1 and 1. The oracle itself is a weighted median of centralised-exchange prices, and the docs say plainly why: it "is robust because it does not depend on Hyperliquid's market data at all." Validators refresh it about every three seconds. Hyperliquid docs

Read that third input again. Your liquidation price on Hyperliquid depends in part on what five other exchanges are doing — venues you did not choose, may not use, and cannot see from inside the app you are trading in. Binance builds its mark from an index too rather than from its own last trade; the specific recipe differs, the principle does not. Blofin

The intent is good, and worth saying plainly: mark price exists so a thin wick on one venue cannot liquidate a position that the wider market never repriced. It protects you from exactly the manipulation you would otherwise be exposed to.

But it has a consequence nobody puts in the onboarding. The price that decides whether you survive is not a price you can see trade. It is a construction. On a calm day it sits within a basis point or two of the last price and the distinction is academic. On the day it matters — a venue dislocating, a book thinning out, an index component wobbling — the construction and the chart come apart, and the one that closes you is the one you were not watching.

We publish the gap every day and call it basis. Today it is −2.8 bps on Hyperliquid's bitcoin and −4.2 on OKX's — perp trading fractionally below the index on both, which is a rounding error. That number is small almost all of the time. It is the number that decides where your liquidation sits, and the fact that it is usually boring is precisely why nobody has a feel for what it does when it is not.

The part that is genuinely unfair, and it is documented

There is a second mechanism underneath liquidation, and it can close a position that is doing nothing wrong at all.

When a liquidation cannot be filled at the bankruptcy price and the insurance fund is exhausted, the venue starts closing profitable traders on the opposite side to make the book balance. It is called auto-deleveraging, and it is a documented, deliberate backstop rather than a malfunction. Gate

Read the ranking. Hyperliquid orders ADL candidates by mark price divided by entry price, multiplied by notional over account value — which is profit multiplied by leverage. The more right you were, and the harder you pressed it, the earlier you get closed. Being correct moves you up the queue.

To be fair to Hyperliquid, it is designed to be rare — the first platform-wide ADL is reported as coming more than two years after launch, in October 2025. Rare is not the same as absent, and "rare" is doing the same work here as "usually boring" did above.

Here is where it lands, and why it belongs in a newsletter about cost and records. An ADL close does not look like anything in your history. There is no signal that fired, no stop you set, no decision you made. Six months later you are looking at an exit you did not choose, at a price that was not on your chart, triggered by a mechanism that was not in your strategy — and nothing in the export says which of those it was. You will file it under "the strategy stopped working."

None of this is hidden. All of it is in the documentation. It is simply never in the part of the documentation anyone reads before their first trade, and the mechanics only become legible on the day they cost you.

Also today

Open interest climbed on every venue while price fell on everything. Bitcoin is at $62,741, down 1.87% in twenty-four hours, and bitcoin open interest rose on all four venues — Bybit +14.7% and OKX +10.4% being the notable ones. Ether the same shape: price down 1.30%, open interest up everywhere, OKX +8.5%. Solana is the exception, down on both.

Positions being added into a falling market is the condition the section above describes. More leverage against a lower price is more of the book sitting nearer its liquidation level, and that is the state in which the difference between mark and last stops being academic.

Bitcoin funding disagrees by twelve and a half points. Hyperliquid and OKX are both pinned at exactly 10.95% — the 0.01%-per-8h baseline again, now on two venues at once — while Bybit has gone negative at −1.73%, having moved 12.7 points in a day. So longs pay on three venues and are paid on the fourth, for the same asset, at the same moment.

The tape

Price and 24h change

	BTC	ETH	SOL
Mark	`$62,741`	`$1,871`	`$75.32`
24h	`−1.87%`	`−1.30%`	`−1.58%`

Funding, per 8h and annualised

	BTC	ETH	SOL
Hyperliquid	`+0.0100%` · `10.95%`	`+0.0100%` · `10.91%`	`+0.0035%` · `3.80%`
Binance	`+0.0009%` · `0.95%`	`−0.0009%` · `−0.97%`	`−0.0054%` · `−5.88%`
OKX	`+0.0100%` · `10.95%`	`+0.0097%` · `10.57%`	`+0.0031%` · `3.40%`
Bybit	`−0.0016%` · `−1.73%`	`+0.0018%` · `2.02%`	`−0.0023%` · `−2.56%`

Hyperliquid settles hourly; its rate is shown multiplied by eight so the column compares. Binance and Bybit are read from CoinGecko rather than directly, and their interval is assumed to be 8h — see the venue note below.

Open interest, basis, and 24h change

	BTC	ETH	SOL
OI, Hyperliquid	`$2.70B` `+5.1%`	`$1.66B` `+2.8%`	`$0.38B` `−0.4%`
OI, Binance	`$7.16B` `+2.4%`	`$4.50B` `+1.5%`	`$0.64B` `−0.9%`
OI, OKX	`$2.28B` `+10.4%`	`$1.40B` `+8.5%`	`$0.22B` `−2.6%`
OI, Bybit	`$4.29B` `+14.7%`	`$1.44B` `+3.8%`	`$0.47B` `−2.1%`
Basis, Hyperliquid	`−2.8 bps`	`−4.8 bps`	`−4.6 bps`
Basis, OKX	`−4.2 bps`	`−4.6 bps`	`−6.6 bps`

Basis is the gap between the perp's mark and the index it is tethered to — the same gap the lead is about. Captured 2026-08-14 14:43 UTC · raw snapshot

Venue watch. Binance returns HTTP 451 and Bybit HTTP 403 to direct requests from this location — geographic blocks, not outages. Their columns come from CoinGecko and are marked as such, so the coverage gate records this issue as degraded: two venues first-party, two second-hand.

One question

Do you know, without looking it up, which price your venue liquidates you on — and where to find it right now? Reply and say. I will publish what comes back, without names.

The cost corner

The venue's own documentation is the only place its actual behaviour is written down, and it is written for developers rather than for you.

The Founder No. 01 — Your exchange is deleting your trading history

Not investment advice. Nothing here predicts anything. Numbers are pulled from venue APIs at the time stamped above and may be revised. Third-party figures are linked to their sources.

Written by Joe DeFeo from inside a live systematic book.