← All articles

Why Fixed Block Times Are the Best Deal LPs Already Have

@0x3hu79
@0x3hu79
April 10, 2026 · MEV Research
#LVR#AMM#liquidity-providers#block-time#MEV#arbitrage

If you provide liquidity on an AMM like Uniswap, you lose money to arbitrageurs every block. Between blocks, the real price moves on centralized exchanges while your AMM quote stays frozen. When the next block arrives, an arbitrageur trades against your stale quote to pocket the difference. This cost - the gap between what your portfolio is worth and what it would be worth if you could continuously rebalance - is called Loss-Versus-Rebalancing (LVR).

LVR is the fundamental tax on passive liquidity provision. It’s also the upstream source of value that funds CEX-DEX arbitrage, which flows to searchers, then to builders, then to proposers. Understanding LVR means understanding the economics that the entire block building pipeline is built on.

The paper by Alex Nezlobin and Martin Tassy provides the first analytical LVR formulas for constant block times - the regime that Ethereum PoS and most modern blockchains actually operate in. The results are both practically useful and conceptually important. Here’s what they found.

The Key Question: Does Block Schedule Matter?

Prior work on LVR (Milionis et al., 2023) derived formulas for Poisson-distributed block times - random arrivals, characteristic of proof-of-work chains like pre-Merge Ethereum. But modern chains produce blocks on fixed schedules: every 12 seconds (Ethereum PoS), every 400ms (Solana), every 2 seconds (many rollups).

Does this difference matter? If your average block time is 12 seconds, does it matter whether blocks arrive exactly every 12 seconds or randomly with a 12-second average?

The answer is yes - and the direction is unambiguous.

Result 1: Constant Block Times Minimize LP Losses

Among all possible block time distributions with the same average block time, constant (deterministic) intervals uniquely minimize per-block LVR. This is proven formally, not just observed in simulation.

The intuition is elegant. LVR is convex in block gap length - longer gaps are disproportionately costly because price drift (and therefore arbitrage opportunity) grows with the square of the time gap. Random block schedules create variance in gap lengths: sometimes short gaps (small loss), sometimes long gaps (large loss). Because the cost function is convex, the average loss under variable gaps exceeds the loss at the constant gap, by Jensen’s inequality.

A simple analogy: if you’re penalized proportionally to the square of your lateness, being consistently 5 minutes late costs you 25 units. Being randomly 0 or 10 minutes late (same average) costs you (0 + 100)/2 = 50 units - twice as much. Variance in timing hurts when the cost is convex.

Ethereum’s PoS design - blocks at fixed 12-second intervals - is already giving LPs the best possible deal for that average block time. No schedule redesign can improve on it.

Result 2: How Much Better Are Fixed Blocks?

Comparing the constant-block LVR formula against the Poisson formula (same average block time), fixed blocks reduce per-block arbitrage losses by up to 17.4% in the fast-block regime.

The improvement depends on the ratio of the AMM’s spread to the per-block volatility:

Block regimePoisson overestimation vs. constant
Very fast blocks (spread >> volatility)~5%
Moderate (spread ≈ volatility)~8%
Slow blocks (volatility >> spread)~15%
Fast-block asymptotic limit17.4%

If you’ve been using Poisson-based LVR estimates for Ethereum PoS analysis, you’ve been overestimating LP losses by 8-17%. The actual cost of providing liquidity on Ethereum is measurably lower than the widely-cited Poisson formulas suggest.

Result 3: Frequency vs. Magnitude - A Clean Decomposition

The paper introduces a decomposition that works for any block distribution:

Expected arbitrage loss per block = P_trade × LVR_bar

Where P_trade is the probability that a given block contains an arbitrage trade, and LVR_bar is the average loss when an arbitrage occurs.

A surprising finding: P_trade is approximately universal. In the relevant regime (small per-block price volatility, which covers most practical block times), the probability of arbitrage occurring in a block is essentially the same whether blocks are Poisson, constant, uniform, or any other distribution.

This means the difference between block distributions comes entirely from LVR_bar - the magnitude of each arbitrage event, not how often arbitrage happens. Constant blocks don’t prevent arbitrage from occurring; they just make each occurrence less costly.

The Practical Formula

For constant block times, the expected per-block arbitrage loss is:

ARB ≈ ℓ × σ_b² / (2 + 1.7164 × γ/σ_b)

Where:

Monte Carlo simulations with a billion paths confirm this formula is practically exact - errors below 0.01% for typical market parameters. This makes it directly usable for estimating per-pool, per-block arbitrage costs on any constant-block-time chain.

What This Means for Shorter Block Times and Sub-Slots

With active discussion around shorter slots (EIP-7782 proposing 6 seconds) and sub-slot execution (TOOL’s 1-second sub-slots, discussed at the Blockspace Forum Workshop), a natural question is: how much does halving the block time help LPs?

The framework gives a precise answer: halving the block time reduces time-normalized LVR by approximately 29%, not 50%.

The math: per-block LVR scales roughly as σ_b² (which is proportional to block time t). Halving t halves per-block LVR. But there are now twice as many blocks per unit time, so the per-unit-time LVR scales as √t. Halving t gives a √2 ≈ 1.41x improvement, or about 29% reduction.

This means sub-slots help LPs, but with diminishing returns:

Each halving delivers the same proportional improvement, but the absolute gain shrinks. The first step from 12s to 6s captures the most value.

This is consistent with the TOOL team’s research showing that shorter sub-slots increase trading volume without dramatically changing individual trader PnL. The paper’s framework explains why: shorter blocks reduce per-event losses (drawing in more LP capital and more trading), but they don’t change the fundamental frequency of arbitrage opportunities per unit of real time.

The Bigger Picture: LVR Funds the Pipeline

LVR isn’t just an LP concern - it’s the upstream economic force that powers the entire block building pipeline:

  1. LPs lose to arbitrageurs through stale AMM quotes (LVR)
  2. Arbitrageurs capture this value through CEX-DEX trades
  3. Searchers package these arbitrage opportunities into bundles
  4. Builders include bundles in blocks and bid for the right to propose
  5. Proposers collect the winning bid

Every analysis of block building economics - from the blockspace market paper’s exclusive order flow dynamics to the JIT liquidity study’s searcher-builder fee pass-through - is ultimately downstream of LVR. The value that builders compete for originates in LP losses.

This paper’s finding that constant blocks reduce LVR by 8-17% compared to Poisson means the MEV pool itself is smaller than Poisson-based estimates suggest. And if sub-slots further reduce LVR, the total extractable value from CEX-DEX arbitrage shrinks - which has cascading effects on builder revenue, relay economics, and the entire blockspace market.

In an encrypted mempool world, where sandwich attacks are eliminated and arbitrage becomes the dominant form of MEV, LVR becomes even more central to the economics. Understanding how block time design affects LVR is understanding how protocol choices shape the revenue that sustains block production infrastructure.

For Block Builders

LVR quantifies your upstream supply. The CEX-DEX arbitrage bundles that are a major source of block value are funded by LP losses. This paper’s formula lets you estimate per-pool, per-block arbitrage revenue for any Uniswap V3 pool - useful for modeling which pools generate the most builder-relevant flow.

The 17.4% overestimation matters for revenue projections. If internal models use Poisson-based LVR estimates, they overstate the available MEV from arbitrage on Ethereum PoS by 8-17%. Correcting for constant block times gives more accurate revenue forecasts.

Sub-slot economics follow directly. If sub-slots reduce per-event LVR (as this paper predicts), arbitrage bundles become individually less valuable but more frequent. Builders who can efficiently process high-frequency, lower-value bundles have an advantage. The builder that wins in a 1-second sub-slot world looks different from the builder that wins in a 12-second slot world.

Sources

Original paper:

Foundational LVR work:

Related Rawcall articles: