TimeBoost Auction Mechanism

• TimeBoost Auction is a hybrid mechanism that combines time-based ordering with explicit bidding to allocate execution priority.
• It employs continuous and discrete designs, using mathematically defined scores to rank transactions and manage latency races.
• Empirical analyses reveal that while it has potential for revenue and welfare gains, practical deployments face challenges like centralization and unreliable execution.

A TimeBoost Auction is a transaction sequencing mechanism that hybridizes temporal priority with explicit bidding, enabling participants to acquire execution priority by bidding for time advantages. Deployed in blockchain rollup sequencers and explored in electronic market microstructure, the TimeBoost paradigm systematically quantifies and prices low-latency access, replacing or supplementing pure first-come-first-served (FCFS) ordering. Its formal implementation spans continuous or discrete time, includes both explicit all-pay time-buying and periodic slot (express lane) auctions, and is designed to discipline latency races, internalize MEV (miner-extractable value) revenue, and theoretically enhance fairness and efficiency. The operational and economic performance of TimeBoost has been evaluated in both theoretical and empirical regimes, notably with the large-scale deployment and subsequent analysis on Arbitrum (Messias et al., 26 Sep 2025), and through comparative studies in financial markets (Mastrolia et al., 16 May 2024).

1. TimeBoost Auction Design: Protocols and Scoring

The TimeBoost mechanism comprises several architectures, distinguished by whether time-buying is continuous (“hybrid time–price ordering”) or discretized into periodic slot auctions. The common feature is an explicit mapping from bids and timestamps to a scalar “score,” which governs sequencing:

• Continuous TimeBoost Ordering (Mamageishvili et al., 2023, Schlegel, 2023): Each transaction arrives at time $t_i$ with bid $b_i$, and is assigned a score

$S(t_i, b_i) = \pi(b_i) - t_i,$

where $\pi(\cdot)$ is a strictly increasing, concave function with $\pi(0) = 0$ and $\lim_{b \to \infty} \pi(b) = g$ for some time window $g > 0$. Transactions are sequenced in descending order of $S$; thus, higher bids functionally reduce the effective timestamp.

• Discrete Slot/Express Lane Auction (Messias et al., 26 Sep 2025, Mamageishvili et al., 23 Nov 2025): Time is segmented into rounds (e.g., 60 seconds). At the start of each round, a sealed-bid, second-price (Vickrey) auction is conducted. The highest bidder becomes the “express-lane controller” for the next interval, gaining a 200 ms latency advantage—submissions bypass a network delay imposed on ordinary transactions. Bidders submit signed commitments off-chain, with the winning bid and controller set on-chain before execution.
• Periodic Call Auction with Time Penalties (Mastrolia et al., 16 May 2024): In market microstructure, the TimeBoost auction is realized as a periodic batch auction: traders can submit at any time within an interval $[0, T]$, with transaction fees increasing in submission time (e.g., linear or quadratic schedule). Clearing time may be randomized to discipline last-minute submission incentives.

2. Mathematical Formalism: Bidding, Pricing, and Equilibrium

Bidding and Scoring

• All-Pay Time Buying (Mamageishvili et al., 2023, Schlegel, 2023): In the all-pay formulation, bidder $i$ purchases time boost $\pi_i \geq 0$ at cost $F_i = c \pi_i / g$, yielding an adjusted timestamp $t_i - \pi_i$. The equilibrium strategy in the symmetric latency case (i.i.d. values $v_i \sim \mathrm{Uniform}[0,1]$) is

$\pi(v) = \frac{g}{2c} v^2,$

with payoff $U_i = v_i \Pr[S_i \leq S_j] - (c/g) \pi_i$. When asymmetric physical latency gaps $\Delta$ exist, the boost can offset the latency up to the equilibrium threshold $u = \sqrt{(c/g) \Delta}$.

• Sealed-Bid Auction Pricing (Messias et al., 26 Sep 2025, Mamageishvili et al., 23 Nov 2025): In each periodic round, bids are ordered $b_{(1)} \geq b_{(2)} \geq \dots$. The winner is $i^* = \arg\max_i b_i$, and pays the clearing price $p = b_{(2)}$ (or the reserve $r$ if only one bid meets the threshold).
• Batch Auction with Time Fees (Mastrolia et al., 16 May 2024): Bidders incur periodic auction fees $\xi(t)$ rising in submission time $t$, with typical schedules $\xi_\ell(t) = a t$ or $\xi_q(t) = a t^2$. This induces strategic timing behavior.

Equilibrium Analysis

• Continuous-Time Equilibrium (Schlegel, 2023):
  • Welfare gap, revenue, and win probability for high-value bidder are all explicit functions of mechanism parameters.
• Discrete-Time Slot Auction (Messias et al., 26 Sep 2025): The sealed-bid auction is weakly dominant-strategy truthful (second-price format).
• Market Microstructure (Mastrolia et al., 16 May 2024): In absence of time-randomization or penalizing fees, the Nash equilibrium is last-moment arrival (τ*=T) by strategic traders, distorting the clearing price and efficiency. Randomization or convex time fees enforce earlier entry.

3. Empirical Evaluation and Performance Metrics

Recent empirical work on Arbitrum’s Timeboost system (Messias et al., 26 Sep 2025, Mamageishvili et al., 23 Nov 2025) provides large-scale evaluation:

• Participation and Centralization (Messias et al., 26 Sep 2025): Over 151,423 auctions and 11.6 million fast-lane transactions, two entities (Selini Capital, Wintermute) controlled >90% of express lane rounds, despite theoretical rotation. The system thus failed to achieve effective decentralization.
• Auction Dynamics and Revenue:

Median bid ratio $\rho = b_{(1)}/b_{(2)}$ stabilized near 1.65 post mid-May. Both clearing prices and participation declined over time; DAO revenues fell from >0.02 ETH/round to <0.005 ETH/round. Total DAO revenue for the window was 1,090.706 ETH.

• Transaction-Level Outcomes:

21.75% of express lane transactions reverted—in contrast, private relay-based MEV bundles nearly always succeeded post-acceptance. Revert rates for controllers ranged from 28% to 49%. Median normalized position was 45.6% of the block, indicating that strict temporal priority did not guarantee block-leading execution slots.

• Spam Mitigation:

The system did not reduce spam. Express lane facilitated large-scale reverted transaction volume, with 22% of such transactions failing. This outcome contradicts Timeboost’s stated goal of spam-resistance.

• Secondary Markets:

Sub-auction intermediaries (e.g., Kairos) attempting to resell fast-lane rights operated at a loss (e.g., Kairos: ~78k USD fees, ~12k USD searcher profits, 48% revert rate), and exited the market. Secondary markets for per-transaction resale collapsed due to low profitability and execution unreliability.

• Bid–Profit Correlation (Mamageishvili et al., 23 Nov 2025): The Pearson correlation between bids and realized 1-minute markouts was weak (0.17–0.33 depending on bidder), increasing when measured over longer aggregation horizons (e.g., over 30–60 minutes: 0.83–0.88). Immediate post-auction delay (15s gap between auction close and execution) and microstructure noise contribute to poor short-run predictive power.

4. Economic Properties, Incentive Structure, and Welfare Analysis

• Truthful and Efficient Allocation (Theoretical):

The second-price format supports truthful revelation in each auction, and continuous TimeBoost mechanisms ensure that—in absence of latency asymmetry—the highest-valuation bidder secures priority at explicit bid cost (Schlegel, 2023, Mamageishvili et al., 2023).

• Revenue–Efficiency Tradeoff:

Explicit formulas show that TimeBoost can strictly dominate pure batching in both welfare and revenue, provided marginal cost per unit boost ($c$) is low relative to maximal boost ($g$), i.e., $c/g \ll 1$.

• Incentive Structure and Strategic Behavior:

In practice, leading searchers adopted trending strategies, setting bids based on recent markouts rather than predicting instantaneous value. The lack of strong pool of competitors led to bid shading and convergence toward minimal revenue-extraction, undermining theoretical revenue guarantees (Messias et al., 26 Sep 2025, Mamageishvili et al., 23 Nov 2025).

• Design Sensitivities:
  • High $g$ allows low-latency competition while limiting latency arms race, but introduces longer finalization for low-bidders.
  • Randomization of clearing time in batch auctions, or increasing fee schedules, can disrupt last-minute gaming and enhance efficiency, with minor increases in mean-squared pricing error (~1–3% improvement, >85% collapse in price-impact, per (Mastrolia et al., 16 May 2024)).

5. Practical Design Variations and Extensions

• Time-Aware Pricing and Reserve Tuning (Heymann et al., 2023): For repeated TimeBoost settings, optimal bid curves can be computed dynamically, adapting reserve prices to age-dependent value curves $k(s)$ per participant. Near-optimal welfare (>99%) can be achieved using simple “shading” heuristics $b_\alpha(t) = \alpha k(t)$ with $\alpha \leq 1$.
• Transaction Fee Schedules and Randomized Deadlines (Mastrolia et al., 16 May 2024): Penalizing submission time via convex fee schedules (e.g., $\xi(t) = a t^2$) is effective in motivating early entry and minimizing strategic price impact. Minor randomization of clearing time suffices to discipline timing without substantial efficiency loss.
• Batch vs. Continuous TimeBoost (Schlegel, 2023): Empirical and analytical comparisons show equivalence between batch auctions ($c/g \simeq 1$) and TimeBoost mechanisms in terms of revenue/welfare when parameters are tuned appropriately.
• Implementation Considerations:

In production systems such as Arbitrum, protocol-level TimeBoost auctions utilitize EIP-712 signed off-chain bids, on-chain verifiability, and strict rotation on minute boundaries, while sequencing mechanics operate via smart contracts and RPC-level APIs (Messias et al., 26 Sep 2025).

6. Limitations, Outcomes, and Broader Implications

• Centralization and Collapsed Competition: The Arbitrum deployment of Timeboost led to centralization, with >90% express-lane control by two entities (Selini and Wintermute), contradicting the rotating priority and fairness objectives of the mechanism (Messias et al., 26 Sep 2025).
• Limited Real-World Value for MEV Searchers: High-value MEV opportunities were clustered toward the end of the block, muting the economic value of minute-level express lane access.
• Spam Not Reduced; Secondary Markets Unsustainable: With 22% of fast-lane transactions reverted, Timeboost failed to deliver on anti-spam goals. Secondary brokers such as Kairos and JetBuilder exited after sustained losses and unreliable execution.
• Revenue Erosion: Protocol income peaked early and declined rapidly as auction competition weakened.
• Inference for Auction Mechanism Design: Performance of ahead-of-time slot auctions suffers from forecast error due to lead time between auction and execution, especially for one-minute windows. Designers are encouraged to (a) reduce post-auction gaps, (b) hybridize with just-in-time mechanisms, or (c) prioritize trend-following signals rather than instantaneous value prediction (Mamageishvili et al., 23 Nov 2025).
• Generalization: The theoretical benefits of TimeBoost (latency arms-race mitigation, truthful bidding, protocol revenue capture) are fundamentally sensitive to equilibrium participation, auction lead time, the volatility of value/extractable profit, and the absence of predictable timing gaps.

7. Summary Table: Mechanism Features and Empirical Outcomes

Aspect	Protocol Design	Field Results (Arbitrum, 2025)
Auction Type	Per-interval, sealed-bid, second-price	Dominated by 2 entities (>90%)
Express Lane Delay Advantage	200 ms	Median normalized block position ≈ 0.46
Bid Format	Off-chain EIP-712 signed bids, on-chain settlement	Median bid ratio 1.52–1.65
Spam Mitigation	200 ms queue delay for non-winners	21.75% revert rate (vs. ≈0% for Flashbots)
Secondary Market Performance	Resale envisioned for express lane rights	Kairos, JetBuilder exited, negative profits
DAO Revenue	1,090.706 ETH (April–July 2025)	Falling revenue, <0.005 ETH/round by July
Fairness/Decentralization	Each round auctioned, theoretical rotation	High centralization, reduced competition

Empirical and theoretical analyses converge on the finding that, while TimeBoost mechanisms are conceptually robust and yield tractable revenue and welfare guarantees in idealized environments, their practical performance is subject to substantial limitations in live deployment. In the presence of concentrated participation, poor alignment between realized and predicted value, and execution timing uncertainty, TimeBoost fails to ensure fairness, decentralization, or effective mitigation of latency and spam (Messias et al., 26 Sep 2025, Mamageishvili et al., 23 Nov 2025).