Citation

Adadurov, A., Barseghyan, S., Chtepine, A., Eloranta, A., Sebyakin, A., Valitov, A. “Open vs. Sealed: Auction Format Choice for Maximal Extractable Value.” nuconstruct.xyz. arXiv:2603.16333v1 [q-fin.TR] (Mar 17, 2026).

Core Contribution

Empirical + theoretical analysis of which auction format maximizes block builder revenue for MEV markets. Uses 2.2 million transactions from three OFAs (Blink, Merkle, MEV Blocker). Derives equilibrium bidding strategies under affiliation.

Key Empirical Regularities

  1. Log-normal MEV value distribution with extreme right-tail concentration

    • Top 1% of transactions → 68% of all MEV revenue
    • Mechanism choice for high-value events dominates overall revenue performance

  2. Heterogeneous competition by MEV type

    • Sandwiches (public mempool observable): >95% bribe ratio (near-full surplus extraction)
    • Naked arbitrage and liquidations: fewer searchers, lower bribe ratios
    • No single format is optimal across all MEV types

  3. Revenue Equivalence Theorem breaks down

    • RET assumes independent private values (IPV); MEV valuations are affiliated (correlated)
    • All searchers observe the same on-chain state and price signals → if it’s valuable to one, it’s valuable to all

Theoretical Results

Affiliation is modeled via Gaussian copula with parameter ρ.

Equilibrium ranking (Milgrom-Weber linkage principle confirmed):

  1. English auction ≥ Second-price sealed-bid ≥ Dutch ≈ First-price sealed-bid ≥ All-pay
  2. The gap: 14–28% revenue advantage for English/SPSB over Dutch/FPSB at moderate affiliation (ρ = 0.5, n = 5 bidders)
  3. At ρ = 0.5, n = 5: English/SPSB yield 17% higher revenue than Dutch/FPSB
  4. Applied to the empirical sample: switching from FPSB to English would have increased revenue by 3.3%–32.0% depending on assumed ρ

$10–18 million in foregone revenue over the sample period from using FPSB instead of English/SPSB formats.

Novel finding: non-monotonicity at high ρ and large n

  • Revenue peaks in the interior of the affiliation parameter space
  • At near-perfect correlation, order-statistic spread collapses → competitive payments decline
  • Implication: increasing competition doesn’t always increase revenue when affiliation is very high

Practical Implications

Current MEV-Boost uses effectively a first-price sealed-bid mechanism (searchers bid their max). The empirical evidence suggests significant foregone revenue. However, moving to English/SPSB auctions introduces latency (sequential bidding) and is hard to implement in real-time block construction.

Builder-level alternatives:

  • Repeated interaction: allows implicit second-price dynamics without formal mechanism
  • Information disclosure: partial information release during the auction increases effective affiliation → pushes toward English-like dynamics
  • SUAVE: designed for more expressive auction formats; enables approaching English-auction outcomes

This Paper vs. MEV Auction Design: Open vs. Sealed, Timeboost, and Kairos Wiki Page

The MEV Auction Design: Open vs. Sealed, Timeboost, and Kairos wiki page covers the high-level findings. This paper page provides the theoretical derivation, affiliation model, and the novel non-monotonicity finding. Key additional detail: the $10–18M figure is the empirical application of the 14–28% gap to the observed 2.2M transaction sample.