[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.] #  @danrobinson Dan Robinson Dan Robinson posts on X about hyperliquid, exchanges, binance, $653m the most. They currently have XXXXXX followers and XX posts still getting attention that total XXXXXX engagements in the last XX hours. ### Engagements: XXXXXX [#](/creator/twitter::273288231/interactions)  - X Week XXXXXXX +1,490% - X Month XXXXXXX +181% - X Months XXXXXXXXX +19% - X Year XXXXXXXXX -XX% ### Mentions: XX [#](/creator/twitter::273288231/posts_active)  - X Week XX +575% - X Month XX +171% - X Months XX +27% - X Year XXX -XXXX% ### Followers: XXXXXX [#](/creator/twitter::273288231/followers)  - X Week XXXXXX +0.51% - X Month XXXXXX +0.83% - X Months XXXXXX +3.70% - X Year XXXXXX +8.20% ### CreatorRank: XXXXXXX [#](/creator/twitter::273288231/influencer_rank)  ### Social Influence **Social category influence** [finance](/list/finance) #3974 [exchanges](/list/exchanges) #322 [cryptocurrencies](/list/cryptocurrencies) XXXX% **Social topic influence** [hyperliquid](/topic/hyperliquid) #20, [exchanges](/topic/exchanges) #44, [binance](/topic/binance) #260, [$653m](/topic/$653m) #2, [$23m](/topic/$23m) #2, [$650m](/topic/$650m) #4, [debt](/topic/debt) #785, [books](/topic/books) #2437, [okx](/topic/okx) 6.9%, [insurance](/topic/insurance) #59 **Top accounts mentioned or mentioned by** [@tarunchitra](/creator/undefined) [@conejocapital](/creator/undefined) [@0xkilgore](/creator/undefined) [@lexnode](/creator/undefined) [@sychype](/creator/undefined) [@stonecoldpat0](/creator/undefined) [@hasufl](/creator/undefined) [@0xnagu](/creator/undefined) [@gakonst](/creator/undefined) [@8bitbytebit](/creator/undefined) [@freecitywarlock](/creator/undefined) [@hyperliquidx](/creator/undefined) [@binance](/creator/undefined) [@tomhschmidt](/creator/undefined) [@giuliorebuffo](/creator/undefined) [@tempo](/creator/undefined) [@mikeippolito](/creator/undefined) [@hydromancerxyz](/creator/undefined) [@sonarxhq](/creator/undefined) [@cobie](/creator/undefined) **Top assets mentioned** [USDC (USDC)](/topic/usdc) ### Top Social Posts Top posts by engagements in the last XX hours "This paper is simply wrong about its central topic: how Hyperliquids ADL works. Tarun is describing a different (much crazier) algorithm which also might explain how he calculated that traders somehow paid $653m to cover a $23m deficit. 🧵" [X Link](https://x.com/danrobinson/status/1998924691365572647) 2025-12-11T01:15Z 80.3K followers, 185.7K engagements "That is just the queue policy with a smaller budget right I am not arguing with you about the budget. I am saying that your "greedy queue allocator" is not how the queue works. It doesn't compute a dollar budget and then take equity from accounts (taking XXX% from each before moving onto the next one) until that budget is recovered. It liquidates based on contracts not entire accounts. Suppose a user with a X ETH short gets liquidated and the budget that is being recovered (computed however you want to compute it) is $1500. The account at the top of the ADL priority queue holds a X ETH long" [X Link](https://x.com/danrobinson/status/1999012411206996219) 2025-12-11T07:04Z 80.3K followers, 8006 engagements "Here is the forest as I see it: X. Your headline number is that traders lost $653m to cover a $23m deficit. That didn't make sense to me so I wondered if your interpretation of the data was mistaken. Happy to dig more into it when you open-source the analysis but if your model of queue-based ADL is that the top N positions all have their equity values zeroed out then I wonder if you might have made the same mistake in your data analysis and interpreted the equity value of the positions being closed as their losses (or something similar to that) * The paper says that Hyperliquid and Binance's" [X Link](https://x.com/danrobinson/status/1999025020140662890) 2025-12-11T07:54Z 80.3K followers, 11.3K engagements "Yes agree that the Hyperliquid algorithm is definitely different from pro rata And I think probably less fair and Sybil-resistant than it. My point is that it is very very different from the Queue algorithm you study in the paper. So I dont think your conclusions from the paper apply to it. And I really want to understand where the $650m number comes from. Again as an accounting identity the losses of the ADLed winners exactly equal the profits of the liquidated losers. So if there was somehow $650m of overshoot losses from the ADLsdue to rounding errors or compounding or whatever you" [X Link](https://x.com/danrobinson/status/1999097411432563060) 2025-12-11T12:42Z 80.3K followers, 8537 engagements "Lets get to one of the big ideas in the paper: that exchanges should take on some platform-level insolvency risk. The paper introduces the concept of severity meaning a parameter an exchange could set where it decides what percentage of bad debt to socialize and how much to keep on its books. The paper suggests that Hyperliquids solvency-at-any-cost regime (severity=1) is suboptimal compared to strategies that dynamically optimize the severity of their loss socialization. But I think severity optimization is a generous name for this. A clearer term would be insolvency optimization" [X Link](https://x.com/danrobinson/status/1999412696182329522) 2025-12-12T09:35Z 80.3K followers, 6575 engagements "My understanding is that most serious exchanges effectively pick severity=1 meaning they socialize losses immediately rather than bearing deficits on their books except in exceptional circumstances like hacks or fraud. And I dont mean to moralize but I think there is a reason the strategy space of calculated automated insolvency management has not been explored before. I think there are many legitimate ways to reduce the harms that socialized losses impose on traders such as insurance funds better loss allocation mechanisms and caps on open interest and leverage. We should be talking about" [X Link](https://x.com/danrobinson/status/1999412699059650847) 2025-12-12T09:35Z 80.3K followers, 6559 engagements "Now maybe Im setting too high a bar for trilemmas. They cant all be the CAP theorem right But the paper literally compares its trilemma to the CAP theorem It says the trilemma echoes famous trilemmas from political science computer science macroeconomics and auction theory which suggests that it is fundamental. I dont understand what they have in common other than the number three" [X Link](https://x.com/danrobinson/status/1999412709964861478) 2025-12-12T09:35Z 80.3K followers, 7154 engagements "TL;DR The paper is not actually talking about ADL but a different socialized loss mechanism. The paper suggests platforms should take on some platform-level insolvency risk which I think is a bad idea. And it justifies this using an ADL trilemma but the proof effectively assumes the controversial part of its conclusion" [X Link](https://x.com/danrobinson/status/1999412712913428558) 2025-12-12T09:35Z 80.3K followers, 7832 engagements "People should stop accusing Tarun of writing his paper to attack Hyperliquid in order to boost his investments or something That is never how he has operated And you should aspire to evaluate ideas by their content rather than based on the person who says them" [X Link](https://x.com/danrobinson/status/1999667888907776405) 2025-12-13T02:29Z 80.3K followers, 88.4K engagements "The Drift code you linked shows that your description of pro-rata does not match what Drift does either. Drift allocates losses proportional to the base asset not proportional to equity. They divide the total losses by the total base asset and then adjust the cumulative_funding_rate on each side by that amount (meaning that like funding rate the loss is applied proportional to the number of contracts held). Do you disagree with that description of how their code works" [X Link](https://x.com/danrobinson/status/1999003163609931802) 2025-12-11T06:27Z 80.3K followers, XXX engagements "The data is actually pretty understandable. There's a big table of all the fills on Hyperliquid. It shows that ADLs are forced sales of positions that are matched against liquidations rather than XXX% haircuts. Tarun's collaborator @ConejoCapital helpfully posted the fills data from 10/10 to GitHub Also there are docs from Hyperliquid Binance OKX HTX that explain how ADLs work" [X Link](https://x.com/danrobinson/status/1999191880022786082) 2025-12-11T18:57Z 80.3K followers, 4313 engagements "@ConejoCapital @MikeIppolito_ @tarunchitra @hydromancerxyz @SonarX_HQ Yes I was impressed with that And from only XX minutes before the flash crash very cool Indeed I would say the data is actually not that available (beyond fills). I've been working on indexing HL for months and still don't have a good indexer for things like USDC transfers" [X Link](https://x.com/danrobinson/status/1999197743273959688) 2025-12-11T19:20Z 80.3K followers, XXX engagements "@tarunchitra @lex_node I definitely don't think you wrote the paper because of your bags and never said that" [X Link](https://x.com/danrobinson/status/1999524572161540311) 2025-12-12T16:59Z 80.3K followers, 1379 engagements "@Sychype @tarunchitra @lex_node I also think ideas should be evaluated on their merits rather than based on who said them" [X Link](https://x.com/danrobinson/status/1999533427414929549) 2025-12-12T17:34Z 80.3K followers, 2203 engagements "@stonecoldpat0 Yep that's basically right And yes it's EVM compatible" [X Link](https://x.com/danrobinson/status/1998423973877707121) 2025-12-09T16:06Z 80.3K followers, XXX engagements "@hasufl Yes hence or fraud" [X Link](https://x.com/danrobinson/status/1999450450232508571) 2025-12-12T12:05Z 80.3K followers, XXX engagements "This is the papers explanation of how Hyperliquid ADL works (shown in chart and in Greek): * Sort all accounts by PNL*leverage * Go down the list * Apply a XXX% haircut to the equity in each account. * Until the $ recovered match the $ lost in the bad liquidations" [X Link](https://x.com/danrobinson/status/1998924693630496887) 2025-12-11T01:15Z 80.3K followers, 28.9K engagements "That does seem unfair Every trader is either wiped out completelytheir whole position confiscatedor untouched. But thats not how Hyperliquids ADL works. Here's how it actually works:" [X Link](https://x.com/danrobinson/status/1998924696293957993) 2025-12-11T01:16Z 80.3K followers, 6489 engagements "* Go through the sorted list * For each account forcibly close the position at the last mark price. * Until the # of ADLed contracts matches the # of insolvently liquidated contracts. ADLed traders aren't wiped out. Their losses are proportional to number of contracts held" [X Link](https://x.com/danrobinson/status/1998924698328117468) 2025-12-11T01:16Z 80.3K followers, 6111 engagements "This paper has gotten a lot of attention both positive and negative but I haven't seen someone point this out. I don't really mind reading difficult papers though I try not to write them. But I think this is a good reason to try to make your papers easier to understand" [X Link](https://x.com/danrobinson/status/1998924705701785949) 2025-12-11T01:16Z 80.3K followers, 11.2K engagements "When I pointed out problems with Tarun's understanding of Hyperliquid he challenged me to read the rest of his paper suggesting I was too dumb to understand it. I don't like obscurantism or gatekeeping. So I did what he asked. Here's what I think: * The paper is wrong about what ADL is even more fundamentally than what I pointed out yesterday. * The paper proposes that exchanges should sometimes take on platform-level insolvency risk which I think is not a good idea. * And the core ADL trilemma is basically a restatement of its own assumptions including an explicit assumption that insurance" [X Link](https://x.com/danrobinson/status/1999412689626628232) 2025-12-12T09:35Z 80.3K followers, 176.8K engagements "First the paper is wrong about what ADL is. It says insolvent positions are sold at a loss which then creates dollar-denominated bad debt that the exchange has to socialize. This is one way to do socialized loss (and some exchanges do it) but its not ADL. And its not how Hyperliquid Binance OKX HTX or BitMEX work. In ADL insolvent positions arent sold on the orderbook. They are closed directly against winning long positions at the time they are liquidated. That was the core innovation of ADLavoiding the slippage caused by liquidating through market orders. As a result there is no concept of" [X Link](https://x.com/danrobinson/status/1999412692575007141) 2025-12-12T09:35Z 80.3K followers, 8337 engagements "Indeed why risk insolvency Why not try other solutions like insurance funds That takes us to the ADL trilemma" the papers core theoretical result. This roughly says that there is a fundamental tradeoff between minimizing insolvency risk preserving fairness and capturing revenue. This suggests that insurance alone will be insufficient to cover tail risks. But how can you prove from first principles that exchange revenue will never be enough to support an insurance fund that can ensure both solvency and fairness" [X Link](https://x.com/danrobinson/status/1999412702154768712) 2025-12-12T09:35Z 80.3K followers, 5686 engagements "Tarun has not responded to my latest points but he RTed this post from @0xnagu defending his paper. So Ill respond to it. No deep understanding of perp mechanics required and no math beyond elementary arithmetic and some logic. I think most of my followers should be able to follow along with a little effort The post is responding to my observation that the papers proof of the ADL trilemma is circular. (Dont get distracted by the part responding to Toly; Toly is talking about an unrelated idea.) Here is the papers informal statement of the trilemma between solvency fairness and revenue:" [X Link](https://x.com/danrobinson/status/2000316496866853148) 2025-12-14T21:26Z 80.3K followers, 31.1K engagements
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]
@danrobinson Dan Robinson
Dan Robinson posts on X about hyperliquid, exchanges, binance, $653m the most. They currently have XXXXXX followers and XX posts still getting attention that total XXXXXX engagements in the last XX hours.
Engagements: XXXXXX #
- X Week XXXXXXX +1,490%
- X Month XXXXXXX +181%
- X Months XXXXXXXXX +19%
- X Year XXXXXXXXX -XX%
Mentions: XX #
- X Week XX +575%
- X Month XX +171%
- X Months XX +27%
- X Year XXX -XXXX%
Followers: XXXXXX #
- X Week XXXXXX +0.51%
- X Month XXXXXX +0.83%
- X Months XXXXXX +3.70%
- X Year XXXXXX +8.20%
CreatorRank: XXXXXXX #
Social Influence
Social category influence finance #3974 exchanges #322 cryptocurrencies XXXX%
Social topic influence hyperliquid #20, exchanges #44, binance #260, $653m #2, $23m #2, $650m #4, debt #785, books #2437, okx 6.9%, insurance #59
Top accounts mentioned or mentioned by @tarunchitra @conejocapital @0xkilgore @lexnode @sychype @stonecoldpat0 @hasufl @0xnagu @gakonst @8bitbytebit @freecitywarlock @hyperliquidx @binance @tomhschmidt @giuliorebuffo @tempo @mikeippolito @hydromancerxyz @sonarxhq @cobie
Top assets mentioned USDC (USDC)
Top Social Posts
Top posts by engagements in the last XX hours
"This paper is simply wrong about its central topic: how Hyperliquids ADL works. Tarun is describing a different (much crazier) algorithm which also might explain how he calculated that traders somehow paid $653m to cover a $23m deficit. 🧵"
X Link 2025-12-11T01:15Z 80.3K followers, 185.7K engagements
"That is just the queue policy with a smaller budget right I am not arguing with you about the budget. I am saying that your "greedy queue allocator" is not how the queue works. It doesn't compute a dollar budget and then take equity from accounts (taking XXX% from each before moving onto the next one) until that budget is recovered. It liquidates based on contracts not entire accounts. Suppose a user with a X ETH short gets liquidated and the budget that is being recovered (computed however you want to compute it) is $1500. The account at the top of the ADL priority queue holds a X ETH long"
X Link 2025-12-11T07:04Z 80.3K followers, 8006 engagements
"Here is the forest as I see it: X. Your headline number is that traders lost $653m to cover a $23m deficit. That didn't make sense to me so I wondered if your interpretation of the data was mistaken. Happy to dig more into it when you open-source the analysis but if your model of queue-based ADL is that the top N positions all have their equity values zeroed out then I wonder if you might have made the same mistake in your data analysis and interpreted the equity value of the positions being closed as their losses (or something similar to that) * The paper says that Hyperliquid and Binance's"
X Link 2025-12-11T07:54Z 80.3K followers, 11.3K engagements
"Yes agree that the Hyperliquid algorithm is definitely different from pro rata And I think probably less fair and Sybil-resistant than it. My point is that it is very very different from the Queue algorithm you study in the paper. So I dont think your conclusions from the paper apply to it. And I really want to understand where the $650m number comes from. Again as an accounting identity the losses of the ADLed winners exactly equal the profits of the liquidated losers. So if there was somehow $650m of overshoot losses from the ADLsdue to rounding errors or compounding or whatever you"
X Link 2025-12-11T12:42Z 80.3K followers, 8537 engagements
"Lets get to one of the big ideas in the paper: that exchanges should take on some platform-level insolvency risk. The paper introduces the concept of severity meaning a parameter an exchange could set where it decides what percentage of bad debt to socialize and how much to keep on its books. The paper suggests that Hyperliquids solvency-at-any-cost regime (severity=1) is suboptimal compared to strategies that dynamically optimize the severity of their loss socialization. But I think severity optimization is a generous name for this. A clearer term would be insolvency optimization"
X Link 2025-12-12T09:35Z 80.3K followers, 6575 engagements
"My understanding is that most serious exchanges effectively pick severity=1 meaning they socialize losses immediately rather than bearing deficits on their books except in exceptional circumstances like hacks or fraud. And I dont mean to moralize but I think there is a reason the strategy space of calculated automated insolvency management has not been explored before. I think there are many legitimate ways to reduce the harms that socialized losses impose on traders such as insurance funds better loss allocation mechanisms and caps on open interest and leverage. We should be talking about"
X Link 2025-12-12T09:35Z 80.3K followers, 6559 engagements
"Now maybe Im setting too high a bar for trilemmas. They cant all be the CAP theorem right But the paper literally compares its trilemma to the CAP theorem It says the trilemma echoes famous trilemmas from political science computer science macroeconomics and auction theory which suggests that it is fundamental. I dont understand what they have in common other than the number three"
X Link 2025-12-12T09:35Z 80.3K followers, 7154 engagements
"TL;DR The paper is not actually talking about ADL but a different socialized loss mechanism. The paper suggests platforms should take on some platform-level insolvency risk which I think is a bad idea. And it justifies this using an ADL trilemma but the proof effectively assumes the controversial part of its conclusion"
X Link 2025-12-12T09:35Z 80.3K followers, 7832 engagements
"People should stop accusing Tarun of writing his paper to attack Hyperliquid in order to boost his investments or something That is never how he has operated And you should aspire to evaluate ideas by their content rather than based on the person who says them"
X Link 2025-12-13T02:29Z 80.3K followers, 88.4K engagements
"The Drift code you linked shows that your description of pro-rata does not match what Drift does either. Drift allocates losses proportional to the base asset not proportional to equity. They divide the total losses by the total base asset and then adjust the cumulative_funding_rate on each side by that amount (meaning that like funding rate the loss is applied proportional to the number of contracts held). Do you disagree with that description of how their code works"
X Link 2025-12-11T06:27Z 80.3K followers, XXX engagements
"The data is actually pretty understandable. There's a big table of all the fills on Hyperliquid. It shows that ADLs are forced sales of positions that are matched against liquidations rather than XXX% haircuts. Tarun's collaborator @ConejoCapital helpfully posted the fills data from 10/10 to GitHub Also there are docs from Hyperliquid Binance OKX HTX that explain how ADLs work"
X Link 2025-12-11T18:57Z 80.3K followers, 4313 engagements
"@ConejoCapital @MikeIppolito_ @tarunchitra @hydromancerxyz @SonarX_HQ Yes I was impressed with that And from only XX minutes before the flash crash very cool Indeed I would say the data is actually not that available (beyond fills). I've been working on indexing HL for months and still don't have a good indexer for things like USDC transfers"
X Link 2025-12-11T19:20Z 80.3K followers, XXX engagements
"@tarunchitra @lex_node I definitely don't think you wrote the paper because of your bags and never said that"
X Link 2025-12-12T16:59Z 80.3K followers, 1379 engagements
"@Sychype @tarunchitra @lex_node I also think ideas should be evaluated on their merits rather than based on who said them"
X Link 2025-12-12T17:34Z 80.3K followers, 2203 engagements
"@stonecoldpat0 Yep that's basically right And yes it's EVM compatible"
X Link 2025-12-09T16:06Z 80.3K followers, XXX engagements
"@hasufl Yes hence or fraud"
X Link 2025-12-12T12:05Z 80.3K followers, XXX engagements
"This is the papers explanation of how Hyperliquid ADL works (shown in chart and in Greek): * Sort all accounts by PNL*leverage * Go down the list * Apply a XXX% haircut to the equity in each account. * Until the $ recovered match the $ lost in the bad liquidations"
X Link 2025-12-11T01:15Z 80.3K followers, 28.9K engagements
"That does seem unfair Every trader is either wiped out completelytheir whole position confiscatedor untouched. But thats not how Hyperliquids ADL works. Here's how it actually works:"
X Link 2025-12-11T01:16Z 80.3K followers, 6489 engagements
"* Go through the sorted list * For each account forcibly close the position at the last mark price. * Until the # of ADLed contracts matches the # of insolvently liquidated contracts. ADLed traders aren't wiped out. Their losses are proportional to number of contracts held"
X Link 2025-12-11T01:16Z 80.3K followers, 6111 engagements
"This paper has gotten a lot of attention both positive and negative but I haven't seen someone point this out. I don't really mind reading difficult papers though I try not to write them. But I think this is a good reason to try to make your papers easier to understand"
X Link 2025-12-11T01:16Z 80.3K followers, 11.2K engagements
"When I pointed out problems with Tarun's understanding of Hyperliquid he challenged me to read the rest of his paper suggesting I was too dumb to understand it. I don't like obscurantism or gatekeeping. So I did what he asked. Here's what I think: * The paper is wrong about what ADL is even more fundamentally than what I pointed out yesterday. * The paper proposes that exchanges should sometimes take on platform-level insolvency risk which I think is not a good idea. * And the core ADL trilemma is basically a restatement of its own assumptions including an explicit assumption that insurance"
X Link 2025-12-12T09:35Z 80.3K followers, 176.8K engagements
"First the paper is wrong about what ADL is. It says insolvent positions are sold at a loss which then creates dollar-denominated bad debt that the exchange has to socialize. This is one way to do socialized loss (and some exchanges do it) but its not ADL. And its not how Hyperliquid Binance OKX HTX or BitMEX work. In ADL insolvent positions arent sold on the orderbook. They are closed directly against winning long positions at the time they are liquidated. That was the core innovation of ADLavoiding the slippage caused by liquidating through market orders. As a result there is no concept of"
X Link 2025-12-12T09:35Z 80.3K followers, 8337 engagements
"Indeed why risk insolvency Why not try other solutions like insurance funds That takes us to the ADL trilemma" the papers core theoretical result. This roughly says that there is a fundamental tradeoff between minimizing insolvency risk preserving fairness and capturing revenue. This suggests that insurance alone will be insufficient to cover tail risks. But how can you prove from first principles that exchange revenue will never be enough to support an insurance fund that can ensure both solvency and fairness"
X Link 2025-12-12T09:35Z 80.3K followers, 5686 engagements
"Tarun has not responded to my latest points but he RTed this post from @0xnagu defending his paper. So Ill respond to it. No deep understanding of perp mechanics required and no math beyond elementary arithmetic and some logic. I think most of my followers should be able to follow along with a little effort The post is responding to my observation that the papers proof of the ADL trilemma is circular. (Dont get distracted by the part responding to Toly; Toly is talking about an unrelated idea.) Here is the papers informal statement of the trilemma between solvency fairness and revenue:"
X Link 2025-12-14T21:26Z 80.3K followers, 31.1K engagements