Summary
MEV auction design determines how much value is captured by the protocol vs. private parties. Comparing open auctions (English, second-price) vs. sealed-bid (FPSB) shows open formats extract 14–28% more revenue. The Timeboost case study on Arbitrum illustrates ahead-of-time auction risks: early deployment created a weak-competition equilibrium where protocol MEV capture fell from 63% to 15%, which the Kairos redesign partially fixed.
Auction Format Comparison
Sealed-Bid First-Price (FPSB)
- Bidders submit bids privately; highest bidder wins and pays their bid
- Problem: bidders shade bids below true value to avoid “winner’s curse”; revenue is lower
- Used by: early MEV-boost iterations, some OFA providers
Second-Price Sealed Bid (SPSB / Vickrey)
- Bidders submit privately; highest bidder wins but pays the second-highest bid
- Incentive-compatible: dominant strategy is to bid true value (no bid shading)
- Higher revenue than FPSB in equilibrium (theoretically equivalent under symmetric bidders; better under asymmetry)
English Auction (Open)
- Ascending price; bidders see current highest bid; auction ends when no one outbids
- Revenue advantage: bidders have less information uncertainty; bid closer to true value
- Problem: reveals information during the auction (latency-sensitive in block production)
Analysis Results (2026 study on sealed vs. open)
- English and SPSB outperform FPSB by 14–28% in MEV revenue capture
- Top 1% of transactions account for ~68% of total MEV revenue — auction design most impacts these high-value events
- Foregone revenue from using FPSB vs. optimal format: estimated $10–18M over the study period
Implication for PBS: MEV-Boost uses a variant of FPSB (builders submit best block; highest wins). Switching to a second-price or open format would capture significantly more value for proposers — but the current architecture makes this difficult to implement without revealing block contents.
Timeboost: Ahead-of-Time Allocation
Timeboost is Arbitrum’s MEV ordering mechanism: users bid in advance for transaction ordering priority within each block. A time boost auction allocates priority rights for a future period.
Original Design (Phase 1)
- Bids submitted in advance (before the relevant period)
- Winner gets guaranteed ordering priority
- Intent: predictable MEV extraction; reduced latency games
The Ahead-of-Time Problem
The original Timeboost design created a weak-competition equilibrium:
- Winner of the advance auction knew they had priority → stopped competing aggressively
- Other bidders knew the winner had priority → reduced their bids
- Result: protocol MEV capture fell from 63% to 15% after adoption
- The winner essentially “locked in” ordering rights cheaply due to lack of real-time competition
This is the classic problem of ahead-of-time commitment in auctions: when you commit to a winner before they know the exact value of what they’re winning, you get worse price discovery than a real-time auction.
Kairos Redesign
Kairos modified Timeboost to address the weak-competition equilibrium:
- Introduced real-time components: the advance auction winner still gets priority, but must compete at execution time for the highest-value opportunities
- Added per-transaction bidding alongside the advance auction
- Result: protocol MEV capture recovered from 15% toward competitive levels
- The advance auction still exists but no longer shields the winner from real-time competition
Lessons
- Ahead-of-time auctions underperform for MEV because MEV value is only known at the time of execution (CEX-DEX arb, liquidations are time-sensitive)
- Commitment reduces competition: once someone has advance rights, others don’t compete as hard
- Hybrid designs (advance auction + real-time bidding) can capture benefits of both predictability and competitive pricing
- Protocol capture ≠ user protection: higher protocol MEV capture doesn’t directly help users; it just changes who captures the MEV
MEV Auction Formats in Production (2026)
| System | Format | Protocol capture | Notes |
|---|---|---|---|
| Ethereum MEV-Boost | FPSB (implicit) | Proposer-side | Relay can adjust bids |
| Arbitrum Timeboost v1 | Ahead-of-time | Fell to 15% | Weak competition problem |
| Arbitrum Kairos | Hybrid | ~40-50% | Real-time + advance |
| CoW Protocol | Batch auction | Solver-side | No ordering MEV |
| SUAVE (future) | Distributed | TBD | Global unified auction |
Open MEV Auction (Sealed vs. Open Relay Bids)
An open MEV auction proposal for PBS would have relays reveal bids progressively rather than in sealed fashion. The challenge: revealing bid amounts lets builders game the auction (submit a bid just above the current highest).
Alternative: multi-relay coordination (discussed at Cannes workshop) — relays share information about bids they’ve seen without revealing specific amounts. This reduces information asymmetry while preventing outright gaming.
Builder Auction and the Latency Race
The current MEV-Boost builder auction has a fundamental latency problem: builders who submit bids later (with more market information) win more often. This creates timing games and geographic advantages.
Options:
- ePBS commit-reveal: builder commits at ~4s, reveals at ~8s; eliminates last-second games
- Sub-slot execution: frequent checkpoints reduce the value of late information
- Sealed bid with early deadline: enforce hard bid deadlines; reduce advantage of delay
Related Pages
- PBS and MEV-Boost — PBS architecture; current auction mechanism
- ePBS: Enshrined Proposer-Builder Separation (EIP-7732) — ePBS: protocol-level auction changes
- Timing Games and Proof-of-Time — Timing games as auction manipulation
- Exclusive Order Flow and the Builder Flywheel — Exclusive flow as a bid-optimization tool
Key Sources
- When Ahead-of-Time Allocation Fails: The Transition to Kairos (2026) — Timeboost case study; 63%→15% protocol capture
- Three Years of PBS (Dec 2025) — sealed vs. open auction comparison; 14–28% uplift; $10–18M foregone; top 1% = 68% of MEV