Episode Summary
Executive Summary: The episode centers on Tarun’s ADL research fallout, the design trade-offs of auto-deleveraging in perp markets, and what October 10 revealed about exchange risk management. Vlad argues for friendlier, more composable exchange design at Lighter, including zero-fee retail access, cross-margin, and Ethereum-native extensibility. The group also debates token-vs-equity tensions in crypto M&A and what should happen when a devco or protocol no longer has a viable standalone future.
Main Topics: ADL research controversy and correction (Priority: 5/5): Tarun explains how his paper on auto-deleveraging triggered backlash from Hyperliquid and Paradigm, then clarifies that his initial estimate overstated losses because he haircut the full position rather than only profits. He argues the directional conclusion remains valid: ADL behavior was more aggressive than ideal and needs better transparency and design. What ADL is and why it matters (Priority: 5/5): The hosts define ADL as a last-resort mechanism used in perp markets when insolvent positions cannot be liquidated profitably. The conversation traces its history from Huobi to BitMEX and Binance, framing ADL as a solvency tool that can impose unfair losses on winning traders. Exchange design trade-offs: fairness, predictability, and solvency (Priority: 5/5): Vlad and Tarun discuss the challenge of designing an ADL algorithm that is both fair and computationally feasible. They contrast greedy queue-based liquidation with more socialized or hybrid approaches, emphasizing that traders need predictability, especially when hedging across venues. Lighter’s product model and competitive strategy (Priority: 4/5): Vlad describes Lighter’s zero-fee retail tier, paid premium tier for lower latency, and a roadmap oriented around Ethereum-native composability, ZK infrastructure, cross-margin, spot, and tokenized positions. The team positions TVL—not volume—as the key North Star metric. Token launch expectations and points farming (Priority: 4/5): The discussion covers pre-TGE distortions, points markets, and the expectation that behavior will change once a token launches. Vlad says Lighter is focused on sustainable growth rather than a day-one price spike, while also trying to curb abusive points farming. Tokens vs equity in crypto M&A (Priority: 4/5): The hosts debate recent acquisitions involving Axelar, Tensor, and Aave-related fee routing, using them to explore whether token holders should have rights when devcos are acquired. The discussion touches on protocol/DAO governance, devco value capture, and whether token holders need stronger protections.
Key Arguments: ADL is a necessary but edge-case mechanism for keeping perp exchanges solvent when liquidations cannot be completed normally. Tarun’s original loss estimate for Hyperliquid was overstated by about 7x, but the paper still identified a real structural issue in how ADL behaved. Transparency matters: decentralized exchanges are better than centralized ones because on-chain ADL events can be observed and analyzed publicly. Predictability is crucial for professional traders; cross-exchange hedges can make isolated ADL behavior dangerous and hard to reason about. Lighter deliberately uses a less aggressive ADL/insurance approach so that traders are protected more often, even if the insurance fund absorbs occasional losses. Zero-fee retail trading can expand market participation and bring in more strategies, while premium fees can still be sustainable for professional users. The future of crypto exchanges likely involves deeper composability with Ethereum, including collateralization, tokenized positions, and shared infrastructure with lending protocols. In crypto M&A, devco acquisitions may be structurally different from token-level acquisitions; token holders often lack clear legal rights, but the governance model may still be flawed. DAOs may need better mechanisms to retain talent and adjudicate change-of-control events, or else value can become split across incompatible claims. TVL is a stronger indicator of genuine trust and capital commitment than volume or open interest, especially during volatile events like October 10.
Data Points: October 10 liquidations: largest number of liquidations in crypto history - Referenced as the major market event that triggered repeated ADL activity across venues ADL trigger count: 20 times in a row (or similar) - Tarun’s early reaction to October 10 on Hyperliquid-style venues Original Hyperliquid overcount estimate: $650 million - Tarun says his paper’s headline estimate was too high after a methodology error Revised estimate: about $100 million - Tarun’s corrected rough estimate of counterfactually lost value from over-aggressive ADL Over-collateralized share of positions: 13% of positions were PL on average - Used to explain why the initial estimate was overstated Overstatement factor: roughly 7x - Tarun’s correction from gross-position haircutting to profit-only haircutting Lighter LLP loss on October 10: $20 million - Vlad says the insurance pool absorbed the hit while traders avoided larger losses Repaid LLP loss: most of the $20 million - Vlad says LLP had largely recovered the loss by the time of the discussion Leverage wipeout today: about $400 million - Mentioned as a separate recent day of market stress Lighter spot TVL in ETH: 2% of TVL - Vlad notes ETH has recently been added as a collateral/spot asset Lighter fee tier: zero fees for retail; paid premium tier for lower latency - Described as the exchange’s two-tier access model Revenue implication for pros: example from $100k/day to $80k/day - Illustrates why professional trading firms may stay even with premium fees
Pivotal Quotes: "If crypto can't fix ADL, if they don't make it so that it's friendly for traditional players, we're just going to have the SDM system..." — Tarun: Tarun frames the existential stakes of ADL design for decentralized perps markets "The goal of decentralized exchanges is to improve upon centralized exchanges." — Tarun: Used to argue that public, observable ADL events are an improvement over opaque centralized handling "TVL is a metric that represents capital that customers trust us with." — Vlad: Vlad explains why Lighter prioritizes TVL over volume or open interest as its core health metric
Implications: Perp exchanges are being judged on more than volume: transparency, fairness, and composability now matter. Better ADL design, cleaner token/equity governance, and Ethereum-native infrastructure may determine which platforms gain durable trust.