Citation
Öz, B., Schlegel, C., Mamageishvili, A. “Just-in-Time Resale in an Ahead-of-Time Auction: An Event Study.” Flashbots / Offchain Labs. arXiv:2603.20175v1 [cs.GT] (Mar 20, 2026).
What Happened with Timeboost
Arbitrum’s Timeboost sells ahead-of-time ordering priority (the right to have transactions boosted ahead of others). Major users — Wintermute and Selini Capital — subsequently adopted Kairos, a just-in-time secondary market system.
This paper is an event study of what happened to the primary Timeboost auction after the JIT secondary market (Kairos) emerged.
Key Finding: Primary Auction Collapsed
After Kairos adoption:
| Metric | Pre-Kairos era | Post-Kairos era |
|---|---|---|
| Bids as % of highest bid | ~62.7% | 14.8% |
| Bidding behavior | Reflects actual CEX-DEX volatility | Decoupled from volatility |
Post-Kairos, demand for timeboosting still exists (CEX-DEX arbitrage demand still responds to price volatility). But observed bids in the primary Timeboost auction no longer reflect this demand.
Interpretation: Kairos captures the JIT resale surplus. The primary auction becomes a de facto fixed cost; the real competition happens in Kairos’s secondary market. Arbitrum (the primary auctioneer) loses the surplus to the Kairos mechanism.
Mechanism: Why the Primary Auction Collapsed
Classic secondary market dynamic:
- Buyers in the primary auction know they can resell in Kairos
- They therefore bid below their true value (planning to profit from resale)
- The primary auction receives low bids even when underlying value is high
- Primary auctioneer (Arbitrum) captures only ~14.8% of value instead of ~62.7%
This is a known problem in ahead-of-time auctions — predicted by theory, now confirmed empirically.
Implications
For Arbitrum
- Primary auction “effectively ceases to function as a meaningful surplus extraction mechanism”
- Total PnL (gross of auction payments) remains similar across regimes → the value is there, just not captured by Arbitrum
- Recommended fixes: modify auction design to reduce primary-secondary gap; dynamic reserve price
For Protocol Design Generally
- Ahead-of-time auctions are vulnerable to JIT secondary markets: any mechanism that sells rights in advance creates resale opportunities
- PBS MEV-Boost currently avoids this partly because the commitment period is very short (one slot ahead)
- Longer-horizon ahead-of-time auctions (e.g., slot auctions for multiple future blocks) would face this problem
For ePBS and Outcome Preconfs
- Preconf systems that sell ordering rights ahead of slot start face the same secondary market risk
- Design of preconf systems must account for potential Kairos-style JIT resale
Connection to Wiki
This paper provides empirical validation for the MEV Auction Design: Open vs. Sealed, Timeboost, and Kairos claim that “Timeboost fell from 63% to 15% protocol capture.” The 63% → 15% figure (slightly different from the 62.7% → 14.8% here) is confirmed, with mechanism explanation.
Related Pages
- MEV Auction Design: Open vs. Sealed, Timeboost, and Kairos — Auction format survey; Timeboost case study
- The L1–L2 Relationship: Settlement, Differentiation, and Native Rollups — L2 sequencer MEV; differentiation vs. L1
- Outcome Pre-Confirmations — Ahead-of-time preconf systems face similar risks