when did ADL happen? on 2025-10-10, hyperliquid performed ADL 35K times across 20K users and 161 coins 99% of these ADL’s occurred during a 5 minute period starting around 21:16 UTC

how do these ADL events line up with price action? nearly all tokens had a chart like the BTC chart shown below: 1) there was a big singular nuke 2) the 5-minute ADL wave occurred mid-nuke 3) the price rebounded afterward

what did this look like for other tokens? all had a similar shape to BTC’s chart, but varying in the speed and extent of price recovery

how would PnL be affected if HL’s ADL’d shorts remained open? for each token there was a narrow window where keeping the short open would have improved PnL but it's likely that most shorts would've stayed open beyond this window, thus decreasing PnL

a few big caveats: - this would be a completely different story if the prices continued to nuke instead of rebound - without ADL the platform would become insolvent, which would also hurt PnL - ADL still had undesirable effects beyond raw PnL, like breaking neutral strategies

any other measures of total ADL impact? hyperliquid ADL’s closed $2.1B of shorts (30% BTC, 20% ETH, 13% SOL, 9% HYPE). 99.9% of ADL happened to shorts rather than longs most closed positions were <$100K in size, but also most of volume came from positions >$100K in size

how did ADL's on HL compare to ADL’s on Binance? not possible to say. binance withholds ADL data and it also doesn’t release complete liquidation data during bursty periods this type of analysis is only possible using the transparent accounting of public blockchains

tldr during the October-2025 market crash, the biggest complaint about HL was that its ADL mechanism robbed the shorts of PnL the opposite was true: HL's ADL increased the raw PnL of most shorts by closing those positions near the price bottom

(thank you to @yq_acc @SoskaKyle @FrankieIsLost @cobie for input)

@notnotstorm @crypto_condom Great data! Though the main complaint was from people running L/S strategies: - getting their shorts ADLed - then being left with a naked long and prices further falling - and finally getting fully liquidated because of that naked long

@0xLouisT @crypto_condom yea definitely. the other effects of ADL were still harmful this data just dispels the most common complaint I saw on CT about HL ADL https://x.com/notnotstorm/stat...

@notnotstorm this is path dependent no? hard to say that ADL wasn’t bad when you don’t know the alternative path (perhaps prices went lower)

@AustinAdams10 you don't know the counterfactual and there's probably some path dependence. however, the price recovery was fastest for tokens where HL is not the dominant venue, which suggests HL ADL was not the dominant factor that caused the rebound

@notnotstorm for my delta neutral account i can assure you this did not happen. the problem is people who are running larger books that include large longs and large shorts. a timing mismatch is going to completely obliterate your account still waiting to touch base with @chameleon_jeff

@seanlippel @chameleon_jeff yea definitely. the other effects of ADL were still harmful this data just dispels the most common complaint I saw on CT about HL ADL https://x.com/notnotstorm/stat...

@notnotstorm I think most complaints were coming from users that had both long and short positions & may have been liquidated as they ended up having a naked long. Agreed for whoever was directional short albeit it could have ended differently if prices did not bounce.

@0xG00gly yea definitely. the other effects of ADL were still harmful

@notnotstorm This is kind of wack argument bc it completely disregard most ppl that are short are systematic so they lost their hedge

@diogenes addressed in the thread https://x.com/notnotstorm/stat...

@notnotstorm Are these calculations based on the actual realized ADL prices or just the mark prices at the time of ADL? Because ADL also includes a component of socialized loss, where the fills are outside mark. I have no idea just how large this was on HL, but it could be significant

@0xdoug for the counterfactual PnL graph it's using median execution prices compared to the executed ADL price there are many assumptions in this analysis. as you say liquidity could have changed a lot between ADL time and later theoretical execution fwiw, looking at these charts for

@notnotstorm It's not a narrative violation it's basic math. Most people on CT can't even do their own uPnL. https://x.com/Tudmotu/status/1...

@Tudmotu it's a bit beyond just basic math because the market could have continued dumping instead of rebounding in that case short PnL would have suffered

@notnotstorm Can you see the ADL price and time in public data? On other exchanges it's possible to be ADLd at T using the oracle price from T - 5 minutes, which has a pretty drastic effect on pnl.

@yenwod_ yea this is all public data. ADL are fills that happen top of block

@notnotstorm the analysis is flawed because many people were long too and they were liquidated when they were ADLed or ended up losing when they could have closed breakeven or even win

@TraderGaucho_ yea the other effects of ADL were still harmful this data just dispels the most common complain I saw on CT about it HL ADL https://x.com/notnotstorm/stat...

@notnotstorm cool data, idk how I feel about it though. I think everyone complained because they didn't know about it and the uncertainty of the timing of execution. if u had just leverage shorts its profitable. But if you had levered positions both long and short it sucked at the time tbh

@fozzydiablo yea definitely. the other effects of ADL were still harmful this data just dispels the most common complain I saw on CT about it HL ADL https://x.com/notnotstorm/stat...