Selected work

Representative Papers

Price Elasticity of Gas Demand on L1 and L2: Evidence from Ethereum and Arbitrum

Jun 11, 2026

This study provides the first empirical identification of the price elasticity of gas demand on Ethereum Mainnet (Layer 1) and Arbitrum One (Layer 2), offering micro-level insights for transaction fee mechanism design and resource pricing. Leveraging on-chain data from 2025–2026, the analysis employs a two-way fixed-effects panel regression combined with an instrumental variables approach to address endogeneity, alongside behavioral clustering and resource-type decomposition. The results reveal an overall elasticity of −0.006 on Layer 1 and −0.036 on Layer 2, with refundable resources on Layer 2 exhibiting a notably higher elasticity of −0.27. Moreover, highly active user clusters demonstrate elasticities up to six times the overall average, underscoring substantial heterogeneity across both resource types and user segments.

0 citationsRead paper

Just-in-Time Resale in an Ahead-of-Time Auction: An Event Study

Mar 20, 2026

This study investigates how the emergence of the instant secondary market Kairos undermines the competitiveness and value capture of Arbitrum’s Timeboost primary auction mechanism. Employing an event study methodology, the analysis integrates bidding behavior, profit attribution, and market dynamics to compare primary and secondary market activity before and after Kairos’ deployment. The research reveals— for the first time—that the instant resale mechanism effectively neutralizes the primary auction’s intended function: the share of total payments captured by the primary auction plummeted from 62.7% to 14.8%, with substantial economic surplus diverted to the secondary market. Building on these findings, the paper proposes novel auction design improvements aimed at recapturing lost value, offering both empirical evidence and theoretical foundations for optimizing on-chain resource allocation mechanisms.

0 citationsRead paper

Timing Games: Probabilistic backrunning and spam

Feb 25, 2026

This study investigates how multiple agents compete for stochastically arriving opportunities under information delays and action costs, with a focus on the resource waste caused by probabilistic backrunning in blockchain systems. The authors formulate the first continuous-time sequential game model with n players that formally captures backrunning as a strategic interaction involving delayed observation and costly actions, and they rigorously characterize its unique symmetric Nash equilibrium. Theoretical analysis reveals that “spam”—the submission of low-value or redundant actions—emerges endogenously as an equilibrium strategy. Furthermore, the work quantifies the worst-case loss in social efficiency, demonstrating that under high competition intensity, the system generates excessive inefficient actions, leading to substantial welfare degradation.

0 citationsRead paper

RAmmStein: Regime Adaptation in Mean-reverting Markets with Stein Thresholds -- Optimal Impulse Control in Concentrated AMMs

Feb 22, 2026

This study addresses the challenge of liquidity provision in concentrated liquidity automated market making, where providers must balance fee revenue against rebalancing costs such as gas fees and slippage, a trade-off poorly handled by existing strategies under dynamic market conditions. The authors formulate liquidity management as an optimal impulse control problem and introduce, for the first time, a partitioning of state space into action and inaction regions. They propose RAmmStein, a deep reinforcement learning approach that incorporates an Ornstein-Uhlenbeck process to model mean-reverting price dynamics and approximately solves the associated Hamilton–Jacobi–Bellman quasi-variational inequality (HJB-QVI) equation for policy optimization in high-dimensional settings. Evaluated on 6.8 million high-frequency observations from Coinbase, RAmmStein achieves a net ROI of 0.72%, reduces rebalancing frequency by 67% compared to greedy strategies, and remains active 88% of the time, substantially enhancing capital efficiency and operational inertia.

0 citationsRead paper

TimeBoost: Do Ahead-of-Time Auctions Work?

Nov 23, 2025

This paper evaluates the predictive validity of bid prices in the TimeBoost pre-auction mechanism for the realized temporal advantage—measured as cumulative fixed-time markout. We employ correlation analysis, time-series modeling, and minute-level dynamic association tests between markout and bid prices. Results show weak correlation between winning bids and realized extraction value, whereas clearing prices (i.e., second-highest bids) exhibit stronger correlation—indicating auction behavior aligning more closely with common-value rather than private-value models. Aggregating data over longer horizons significantly improves correlation, suggesting bidders excel at identifying directional trends rather than precise timing. Furthermore, markout from the prior minute exerts a statistically significant influence on current bidding, reflecting market reliance on recent value signals for dynamic price formation. The core contribution is the first empirical characterization of TimeBoost’s value-discovery mechanism and its inherent limitations in temporal prediction accuracy.

0 citationsRead paper

Recent publications

Latest Papers

Price Elasticity of Gas Demand on L1 and L2: Evidence from Ethereum and Arbitrum

Jun 11, 2026

This study provides the first empirical identification of the price elasticity of gas demand on Ethereum Mainnet (Layer 1) and Arbitrum One (Layer 2), offering micro-level insights for transaction fee mechanism design and resource pricing. Leveraging on-chain data from 2025–2026, the analysis employs a two-way fixed-effects panel regression combined with an instrumental variables approach to address endogeneity, alongside behavioral clustering and resource-type decomposition. The results reveal an overall elasticity of −0.006 on Layer 1 and −0.036 on Layer 2, with refundable resources on Layer 2 exhibiting a notably higher elasticity of −0.27. Moreover, highly active user clusters demonstrate elasticities up to six times the overall average, underscoring substantial heterogeneity across both resource types and user segments.

0 citationsRead paper

Just-in-Time Resale in an Ahead-of-Time Auction: An Event Study

Mar 20, 2026

This study investigates how the emergence of the instant secondary market Kairos undermines the competitiveness and value capture of Arbitrum’s Timeboost primary auction mechanism. Employing an event study methodology, the analysis integrates bidding behavior, profit attribution, and market dynamics to compare primary and secondary market activity before and after Kairos’ deployment. The research reveals— for the first time—that the instant resale mechanism effectively neutralizes the primary auction’s intended function: the share of total payments captured by the primary auction plummeted from 62.7% to 14.8%, with substantial economic surplus diverted to the secondary market. Building on these findings, the paper proposes novel auction design improvements aimed at recapturing lost value, offering both empirical evidence and theoretical foundations for optimizing on-chain resource allocation mechanisms.

0 citationsRead paper

Timing Games: Probabilistic backrunning and spam

Feb 25, 2026

This study investigates how multiple agents compete for stochastically arriving opportunities under information delays and action costs, with a focus on the resource waste caused by probabilistic backrunning in blockchain systems. The authors formulate the first continuous-time sequential game model with n players that formally captures backrunning as a strategic interaction involving delayed observation and costly actions, and they rigorously characterize its unique symmetric Nash equilibrium. Theoretical analysis reveals that “spam”—the submission of low-value or redundant actions—emerges endogenously as an equilibrium strategy. Furthermore, the work quantifies the worst-case loss in social efficiency, demonstrating that under high competition intensity, the system generates excessive inefficient actions, leading to substantial welfare degradation.

0 citationsRead paper

RAmmStein: Regime Adaptation in Mean-reverting Markets with Stein Thresholds -- Optimal Impulse Control in Concentrated AMMs

Feb 22, 2026

This study addresses the challenge of liquidity provision in concentrated liquidity automated market making, where providers must balance fee revenue against rebalancing costs such as gas fees and slippage, a trade-off poorly handled by existing strategies under dynamic market conditions. The authors formulate liquidity management as an optimal impulse control problem and introduce, for the first time, a partitioning of state space into action and inaction regions. They propose RAmmStein, a deep reinforcement learning approach that incorporates an Ornstein-Uhlenbeck process to model mean-reverting price dynamics and approximately solves the associated Hamilton–Jacobi–Bellman quasi-variational inequality (HJB-QVI) equation for policy optimization in high-dimensional settings. Evaluated on 6.8 million high-frequency observations from Coinbase, RAmmStein achieves a net ROI of 0.72%, reduces rebalancing frequency by 67% compared to greedy strategies, and remains active 88% of the time, substantially enhancing capital efficiency and operational inertia.

0 citationsRead paper

TimeBoost: Do Ahead-of-Time Auctions Work?

Nov 23, 2025

This paper evaluates the predictive validity of bid prices in the TimeBoost pre-auction mechanism for the realized temporal advantage—measured as cumulative fixed-time markout. We employ correlation analysis, time-series modeling, and minute-level dynamic association tests between markout and bid prices. Results show weak correlation between winning bids and realized extraction value, whereas clearing prices (i.e., second-highest bids) exhibit stronger correlation—indicating auction behavior aligning more closely with common-value rather than private-value models. Aggregating data over longer horizons significantly improves correlation, suggesting bidders excel at identifying directional trends rather than precise timing. Furthermore, markout from the prior minute exerts a statistically significant influence on current bidding, reflecting market reliance on recent value signals for dynamic price formation. The core contribution is the first empirical characterization of TimeBoost’s value-discovery mechanism and its inherent limitations in temporal prediction accuracy.

0 citationsRead paper