# Optimal Trading Speed on Blockchain

Crypto.world

> Uses Base's tenfold block-speed upgrade as a natural experiment: faster blocks cut priority-fee bids, shrink pool depth and widen spreads, so block-time roadmaps need rethinking.

**arXiv**: 2611.13616

**Published**: 2026-11-10

**Authors**: Ruizhe Jia, Shihao Yu

## AI summary

## Overview

**Research area:** Trading and market microstructure — specifically blockchain market design, automated market maker (AMM) liquidity provision, priority-fee/MEV rent allocation, and Layer 2 execution.

**Technical level:** Advanced. The paper combines a dynamic equilibrium model (Poisson block arrivals, a priority-fee bidding game between two arbitrageurs, an active versus passive liquidity provider problem, and a free-entry condition for passive capital) with a difference-in-differences empirical design on on-chain transaction data.

**Scope (one sentence):** The paper shows, in theory and with evidence from Base's Flashblocks upgrade, that shortening the block interval shifts arbitrage rents from the sequencer to arbitrageurs, makes active liquidity provision profitable enough to cream-skim passive providers, and produces an interior block speed at which equilibrium pool liquidity is highest.

## What This Paper Is About

On a blockchain, trades do not execute as they arrive; they wait for the next block, and the sequencer that builds the block decides their order and usually sells that priority through fees. The block interval is therefore a design parameter that every AMM pool on the chain inherits as its execution clock, and chains are currently competing to shorten it (Base moved from two seconds to 200 milliseconds via Flashblocks on July 16, 2025).

The paper asks how that clock alone — holding every participant's message speed fixed — changes who captures arbitrage profits, whether liquidity providers actively withdraw ahead of toxic flow, and how much liquidity a pool ultimately offers traders. The goal is an equilibrium model of these three margins plus a quasi-experimental test of each prediction around the Base upgrade.

## Key Contributions

1. **A tractable equilibrium model in which block speed is the only varied parameter** and jointly determines the division of arbitrage rents among the sequencer, arbitrageurs, an active liquidity provider, and passive liquidity, with an interior liquidity maximum arising endogenously.
2. **A new mechanism for the allocation of priority rents:** block speed changes whether arbitrageurs "clash" (arrive in the same block and enter a priority auction) or "race" (compete across blocks on arrival order), so faster blocks move the chain toward first come, first served without any participant becoming faster.
3. **An explanation of why active liquidity management emerges on fast chains:** short blocks let a withdrawal execute in an earlier block than the arbitrage trade, so the active provider can dodge arbitrage losses it cannot outbid (the "tragedy of the commons" of Capponi and Jia, 2025) under slow blocks, while still sharing trading fees.
4. **Quasi-experimental evidence on all three predictions** using Base's Flashblocks launch as the treatment, Arbitrum as the comparison chain, and the Uniswap v3 WETH–USDC 5 bp pool matched to Binance order-book snapshots.

## Main Findings

- **Sequencer rents fall, arbitrageur rents rise:** Priority-fee (gas) bids fall most on large arbitrage trades; the decline grows with trade size across the top three size deciles, reaching 73% in the top decile. Smaller trades, which pay only the baseline congestion price, show no change.
- **The prize shrinks, but the cost of capturing it shrinks more:** The blockchain fee falls by 3.26 bps while the raw markout before the fee falls by only 0.53 bps (statistically insignificant), so the net markout rises by 2.73 bps. In dollar terms, gross profit extracted from the pool falls by about 29% on Base while net profit rises by 40%.
- **The active LP becomes genuinely active:** Her rebalancing time falls by about 66% after Flashblocks, while her gas price does not change relative to Arbitrum — she does not bid more for priority, she simply gets earlier blocks.
- **Active management turns profitable:** Her per-dollar profit rises by 0.73 bps, moving from approximately break-even to positive, and she scales up her liquidity capital substantially.
- **Cream-skimming is measurable:** Her share of exposure to toxic flow (trades for which the price gap against her exceeds the fee she earns) falls by 8.93 percentage points more than the pool's toxic-flow share, netting out changes in the mix of flow reaching the pool.
- **Depth falls and trading gets more expensive:** Depth at the current pool price falls by about 19%, the largest single liquidity provider's share of pool liquidity rises by 4.20 percentage points, the effective spread on trades of about 1 ETH widens by 0.07 bps (a 47% increase, larger for larger trades), and trading volume falls by 18%.
- **Faster is not unambiguously better:** The model predicts an interior liquidity maximum in block speed; beyond a threshold, the active provider's income keeps rising while passive capital retreats and total pool liquidity declines. The Base evidence is consistent with being on the downward-sloping side of that curve.
- **Motivating stylized fact:** Around Flashblocks, arbitrageurs' net profit per dollar traded (10-second markout against the Binance mid-quote, net of the blockchain fee) nearly doubled, and the effective spread widened — the opposite of the naive prediction that faster price updating improves liquidity.

## Methodology in Plain English

The model has a single risky asset on an AMM. Blocks arrive as a Poisson process with rate B, so the expected block interval is 1/B, and B is set by the protocol and taken as given. Two arbitrageurs watch the gap between the pool price and an external reference price; when news opens a profitable gap, each sends a transaction that reaches the sequencer after an exponentially distributed latency with rate λ, and each chooses what percentage of the arbitrage prize to bid for priority, without seeing whether its rival has arrived. Liquidity traders arrive as a Poisson process with exponentially distributed deadlines at rate δ, so the fee flow actually served is r̄ · B/(B+δ) — faster blocks mean fewer traders give up waiting. Passive LPs deposit capital P, bear a holding cost k per unit of capital per unit of time, and can enter freely. One active LP has fixed capital q and withdraws or redeposits in full, with the same latency rate λ as the arbitrageurs, so she has no speed advantage — only the block clock on her side. The price gap takes values in {0, M, H} with 0 < M < f < H, where f is the fee threshold; news arrives at rate J, moving 0 → M, and then M → 0 or M → H with equal probability.

On the empirical side, the authors use Base's public Flashblocks launch on July 16, 2025 as a quasi-experiment: the effective block interval fell from two seconds to 200 milliseconds (a tenfold increase in speed) while the priority gas auction ordering rule inside each Flashblock was preserved. Arbitrum, whose block interval did not change over the sample, is the comparison chain; during the sample Arbitrum used first-come, first-served ordering with an auction for an express lane. The data are on-chain transactions in the Uniswap v3 WETH–USDC 5 bp pool on Base, matched to Binance order-book snapshots, from May 16 to September 15, 2025, excluding the July 7–16 transition window and starting the post-launch sample on July 17. A difference-in-differences design then tests the rent-redistribution, cream-skimming, and depth predictions, with arbitrage markouts measured over 10 seconds against the Binance mid-quote and effective spreads computed relative to the pool price just before each trade, excluding the pool fee.

## Why This Matters

**Impact on research.** The paper connects the traditional-market literature on trading speed (Budish et al., 2015; Menkveld and Zoican, 2017; Biais et al., 2011; Du and Zhu, 2017) to AMM microstructure by identifying a speed parameter that participants cannot privately invest in — the block clock — which cleanly separates the effect of speed from the effect of being fast. It also extends the AMM adverse-selection literature (Lehar and Parlour, 2023; Milionis et al., 2023; Capponi and Jia, 2025) and the just-in-time liquidity literature (Wan and Adams, 2022; Capponi et al., 2024) by endogenizing when active provision arises, and it links to the MEV and proposer-builder-separation literature (Daian et al., 2020; Schwarz-Schilling et al., 2023; Bahrani et al., 2024) and to work on the same upgrade by Wu and Öz, 2026.

**Real-world applications.**

- **Chain design:** Protocols choosing a block interval or sub-block cadence (Base at 200 milliseconds, OP Mainnet at 250 milliseconds since September 2025, ZKsync Era at one-second blocks, Ethereum at 12 seconds) face a documented trade-off between serving liquidity traders and enlarging arbitrage rents and spreads.
- **Tokenized equities:** The SEC's temporary conditional exemptions of September 17, 2026 require tokenized U.S. stock AMM pools to be deployed on public, permissionless blockchains, making the chain's execution clock part of the market design for those pools — and the paper argues faster blocks can undermine the passive, long-horizon liquidity supply that equities need (Malinova and Park, 2023).
- **LP strategy and risk management:** The results give quantitative benchmarks for when active repositioning beats passive deposit-and-hold, and show that the gain comes from avoiding toxic flow rather than from winning priority auctions.
- **Sequencer and MEV economics:** The finding that priority bids collapse on large trades (73% in the top decile) informs fee-mechanism and ordering-rule design, including the shift of Arbitrum from FCFS to a priority gas auction in September 2026.

**Industry relevance.** Decentralized exchanges recorded approximately $233.5 billion in spot volume and $553 billion in perpetual futures volume in August 2026, and rollups already carry most of Ethereum's activity (about 15.8 million transactions per day across nine rollups versus 1.7 million on Ethereum in August 2025). For exchanges, sequencers, and market makers operating at that scale, block speed is a first-order economic parameter rather than an engineering detail.

## Future Directions

1. **Where is the interior liquidity maximum, empirically?** The model predicts a unique intermediate block speed that maximizes pool depth, and the Base evidence places the chain past the peak, but the paper does not estimate the location of that peak or map it to observable chain parameters.
2. **Ordering rules versus block speed as separate instruments.** Base pairs 200-millisecond Flashblocks with a priority gas auction, while Arbitrum used first-come, first-served with an express lane during the sample and adopted a priority gas auction in September 2026. Separating the effect of the clock from the effect of the rule is an open design question the paper's framework sets up but does not resolve.
3. **Welfare and cross-market effects.** The model produces a clear redistribution among sequencer, arbitrageurs, active LP, passive LP, and liquidity traders, but the paper does not quantify the aggregate welfare consequences, nor how the 18% volume decline and wider spreads interact with where tokenized equities ultimately list.
4. **Generalizing beyond two arbitrageurs and a single active LP.** The tractability of the model comes from its small player set and stylized gap process ({0, M, H}); whether the clash-versus-race mechanism and the interior liquidity peak survive richer competition, concentrated liquidity ranges, and multi-venue arbitrage is left open.

## Target Audience

Market-microstructure researchers and graduate students working on trading speed, batch auctions, and dealer/market-maker behavior; academic and industry economists studying DeFi, AMM design, and MEV; blockchain protocol designers and sequencer operators choosing block intervals and ordering rules; quantitative market makers and liquidity providers deciding between passive and active strategies on Layer 2 chains; and policy and market-structure staff evaluating tokenized-equity trading venues, who will find the SEC exemption context and the liquidity-quality results directly relevant.

## Authors’ abstract

On blockchain-based exchanges, faster blocks weaken the priority-fee competition among arbitrageurs and shift rents from the block producer to them. Equilibrium liquidity peaks at an interior block speed; beyond a threshold, an active liquidity provider cream-skims and passive capital shrinks. Evidence from the Base Flashblocks upgrade, a tenfold block-speed increase, confirms the predictions. Priority-fee bids decrease sharply and arbitrageurs' net profit rises. The active…

[The full abstract on arXiv](https://arxiv.org/abs/2611.13616)

[Read the original paper](https://arxiv.org/abs/2611.13616)

---

Canonical: https://crypto.world/paper/optimal-trading-speed-on-blockchain-2611.13616
