Bitunix mark vs last on ETH, advanced: which way the gap hurts, drilled on a real 20x liquidation line

Live Bitunix ETHUSDT perp numbers from 4:00 PM ET, Nov 3, 2026: mark 2,290.62, last 2,290.15, index 2,291.48, funding 0.01% per 8h with a +/-0.3% cap. Spot ETH on Bitunix read 2,291.3 at the same time. The perp's 24h range was 2,240.21 to 2,308.3 from a 2,248.39 open, a 68.09 swing or about 3.03%.

The three gaps

Last minus mark = 2,290.15 - 2,290.62 = -0.47. Mark minus index = 2,290.62 - 2,291.48 = -0.86. So the perp traded a little under spot, and last was the lowest of the three.

On Bitunix, mark drives unrealized PnL and liquidation checks. Last drives fills and your candle chart. TP/SL can be set to LAST_PRICE or MARK_PRICE, and the default is the market (last) price unless you change it.

A real liquidation line

Bitunix's own risk guide gives a simple long approximation: liquidation = entry x (1 - 1/leverage + maintenance margin rate). It uses 0.5% maintenance as its example, so that is the rate used here; your real tier may differ and the estimate in your position panel wins.

Long 2 ETH at 2,290.15, 20x. 1/20 = 0.05. Liquidation = 2,290.15 x (1 - 0.05 + 0.005) = 2,290.15 x 0.955 = 2,187.09. Distance from mark = 2,290.62 - 2,187.09 = 103.53, about 4.52% of mark. Distance from last = 103.06. Nearly the same today, but the mark figure is the one that gets checked. Funding at 0.01%: 2 x 2,290.62 x 0.0001 = 0.46 USDT per 8h window while the rate holds. Longs pay when it is positive.

Which direction hurts

Case 1, last below mark (today). In a fast drop, last usually leads and mark lags because it is anchored to index. A last-price stop fires first. For a long, that is the safe ordering: the stop gets you out before mark reaches liquidation.

Case 2, last above mark. This shows up when the perp has been trading rich and the premium fades, or when spot sells off while the perp book is slow to follow. Mark can fall through 2,187.09 while last is still printing above a 2,195 stop (hypothetical). Liquidation wins the race and the stop never fires.

Case 3, a wick on the Bitunix book alone. One large market sell prints last far below the wider market. A last stop fires; mark barely moves; a mark stop stays put.

So the trigger choice depends on what you fear more: a wick shaking you out (mark) or a slower exit in a broad crash (last tends to fire sooner for a long when last leads).

Spacing with real numbers

With liquidation at 2,187.09, a stop at 2,195 leaves 7.91, about 0.35% of mark. A stop at 2,210 leaves 22.91, about 1.0%. On a day when ETH moved 3.03% from low to high, the second spacing is the one that survives Case 2.

Common mistakes

1. Treating the default trigger as neutral. It is last unless you change it.
2. Measuring buffer from the last candle instead of mark.
3. Ignoring the sign of the basis. The same stop behaves differently at a discount and at a premium.
4. Raising leverage without re-running the line. At 50x the same formula leaves only 1.5% between entry and liquidation.
5. Dragging a stop closer to liquidation to dodge a loss, which hands the exit to the liquidation engine.

FAQ

Q: Does mark ever equal last? A: Often within cents on ETH. The gap widens in fast or thin markets, which is exactly when it matters.

Q: Does a mark trigger guarantee my exit price? A: No. The trigger only decides when the order goes out. Bitunix TP/SL orders default to market, so the fill comes from the book.

Q: Is BTC behaving the same way? A: At 4:00 PM ET Bitunix BTCUSDT showed mark 77,855.1, last 77,855 and index 77,877.7, so mark sat 22.6 under index. Same discount pattern, almost no last-mark gap.

Run this drill on ETH first, then BTC, SOL, XRP and DOGE, where thinner books stretch the gap faster. Tonight, open your own position and write down which side of the gap you are on.