Data Revealed: How Much Can MEV Bots Earn from CEX-DEX Arbitrage?

This article is from: Flashbots data analyst danning

Compiled by | Odaily Planet Daily (@OdailyChina); Translated by | Azuma (@azuma_eth)

How much profit can MEV arbitrage bots actually earn from CEX-DEX arbitrage?

Previously, no one could answer this question, but we are excited to announce that a new paper using formal methods for measurement has finally been published (paper link: https://arxiv.org/abs/2507.13023), I will summarize all the core findings of the paper with a series of images and explanations below.

• Profit? Considerable, but not as much as you think;

• Bot strategies vary, but the excess returns of top traders mostly decay within 0.5 ~ 2 seconds;

• The market is becoming increasingly centralized, and so is the field of block builders;

• However, as competition among blockchain builders intensifies, the profit margin of CEX-DEX arbitrage is compressed year by year;

1. Bots are deeply integrated into the block construction process in various ways;

2. The deeper the binding with the block builder, the thinner the "surface" profit (actually transferred to related parties);

3. The smaller the market share of the block builder, the higher the actual profit margin retained by its associated arbitrageurs;

• Even firmly ranking in the top two in the industry, block construction is still a tough business (with paper-thin profits).

In the 1 year and 7 months of data we collected, the performance of 19 leading CEX-DEX arbitrage bots is as follows:

• Total transaction volume reached $241 billion;

• Extracted $233.8 million in profits;

• Retained only $90.1 million in net profit (paid $143.7 million in shares to block builders);

Overall, the average profit margin of CEX-DEX arbitrage is 38.5%.

Based on the market share analysis of arbitrageurs, we confirm that the centralization trend of the MEV market in CEX-DEX has reached a "highly monopolistic" level.

Using the "League of Legends" tier label system proposed by @0x Rezin, we calculated the Binance markouts of arbitrage bots, using weighted averages to define their "total income" before hedging.

The data shows that most CEX-DEX arbitrage signals disappear rapidly within seconds. The peak income can be observed through the median distribution — the best hedging opportunity appears in the 0.5-1.5 second interval.

After deducting the shares paid to block builders, we obtained the upper limit estimate of Bot profits.

So, after combining the profit corrections of arbitrageurs, how exactly are the earnings of the current top three block builders?

Since rsync (currently ranked third) abandoned the "order flow war" in the middle of last year, its market share has significantly plummeted, but what no one noticed is that its profit margin has rapidly rebounded from 5% to 25%, which brings its comprehensive profit margin (arbitrage + block construction) to about 27%.

However, the actual profits of the top two block builders are very limited.

In the 18-month data cycle, beaverbuild (currently ranked first) has a comprehensive profit margin of only 7.92% (including arbitrage income), while Titan (currently ranked second) without proprietary arbitrage has a profit margin of only 5.85%.

Obviously, the opaque "order flow" trading makes this situation even harder to explain.

In addition to the known "block builder + arbitrageur" combinations such as beaverbuild + SCP, rsync + Wintermute, etc., correlation analysis reveals another set of significant exclusive cooperation cases. Observing the 30-day rolling correlation between "Kayle's transaction volume share in Titan's constructed blocks" and "Titan's market share" in the figure below, you can see the clue.

Our core conclusion is that block construction is a low-profit business, and without holding order flow with ultra-high MEV value, there is no opportunity to enter the market today.

In addition, the current block auction mechanism has serious inefficiency problems.On one hand, the subsidy mechanism will squeeze the profits of block builders; on the other hand, exclusive cooperation will fragment the order flow, prolonging the waiting time for transactions to be on-chain.

But the status quo is not unchangeable. The newly launched BuilderNet by Flashbots may be able to break the deadlock and improve the income of block builders.