A 5:1 ratio began circulating in August 2026. It came from a Bitquery study, amplified by CryptoSlate, and it has the clean shape that crypto markets adore. Ethereum builders, the story goes, extract five dollars for every one dollar the network burns. The shape is clean. The ledger behind it is not.
I have spent more than a decade watching Ethereum's settlement layer through the wrong end of a Dune query, and the first rule I learned is that receipts are not profits. A builder's fee recipient can be full to the brim while the builder is still losing money. The second rule is that an untraced transfer is an alibi, not a finding. It can tell you where money appeared, but it cannot tell you where money finally landed.
This is not a hit piece on Bitquery. It is a warning about letting a headline own the methodology. The underlying report is a serious attempt to quantify arbitrage economics, and the exchange-level price dislocations it tracks are real. But once a number enters public discourse, it takes on a forensic status it never earned. So let us do the forensic step the market skipped. Follow the gas, not the narrative.
The Block Auction Has Two Legs
To understand why 5:1 is fragile, you have to understand how Ethereum block production works after The Merge and after Flashbots turned MEV into an industrial market.
Ethereum does not ask searchers to send bundles directly to validators. Searchers scan the mempool and on-chain liquidity for arbitrage opportunities. They find a mispriced DEX pair, or a liquidation that should not exist, and they package their transactions into a bundle. A bundle is a promise: if you include my transactions, you will extract this profit. Searchers then submit those bundles to builders.
Builders are the new middlemen. They collect ordinary user transactions, arbitrage bundles, liquidation packages, and priority fee tips. They assemble an actual block. Then they participate in a block-space auction. The winning builder pays the validator, in Ethereum, for the right to propose the block. That last payment is the second leg, and it is the leg that public ratio charts often forget.
The actors are easy to confuse. The proposer is the validator that gets to propose a block. The builder is the party constructing the block. Under EIP-1559, part of every transaction fee is destroyed by the protocol, and part is a priority fee that can be routed to the fee recipient. In the Flashbots model, the builder typically sets itself as the fee recipient or pays an external recipient chosen by the validator. The builder then includes a transaction near the end of the block that sends ETH to the validator's address.
This is not an obscure technicality. It is the center of the accounting problem.
The Receipt Is Not the Revenue
Imagine a simple arbitrage block. A searcher finds a Uniswap pool trading ETH at a stale price. The searcher routes a large trade through the pool and locks in profit. To secure inclusion, the searcher pays the builder a priority fee or a direct tip.
The builder includes the searcher's bundle. It also includes several other high-fee transactions from regular users. The builder's fee recipient now shows a healthy inflow. A researcher looking at fee-recipient addresses could record that total as builder revenue.
But the builder has not won the block yet. It must pay the proposer. In a competitive auction, that payment can be substantial. Suppose the builder's fee recipient collected 6 ETH from all transactions and bundles. Suppose the builder had to pay 3 ETH to the validator to win the slot. The builder's economic revenue is closer to 3 ETH, not 6 ETH. If a data pipeline ignores that second transfer, it will overstate builder income by 100% on that block.
It gets worse when a builder also pays external relays or uses complex fee recipient logic. The address that receives the first inflow may not be the address that keeps the money. The same block can contain a chain of internal transfers: from searcher to builder, from builder to fee recipient, from fee recipient to proposer. Each transfer is visible on-chain, but only if the study is designed to trace it.
New York Fed Staff Report 1102 offers the correct accounting frame. I have used that frame in my own Dune work, and it should be mandatory in MEV research. Builder profit equals direct payments plus priority fees minus the amount paid to the proposer. That is not an opinion. It is a ledger identity. If a study reports builder revenue without subtracting proposer payments, it is reporting gross block auction revenue, not builder profit.
The Bitquery write-up does not show enough detail for me to confirm whether that subtraction happened. The public headline suggests it did not. A sum of fee-recipient inflows, divided by EIP-1559 burn, produces a number that sounds like a statement about value capture. But it may actually be a statement about the cost of buying block space before that space is sold to validators.
I cannot call this fraud. I can call it incomplete chain of custody. In my 2017 audits of ICO smart contracts, I learned that code can look clean until you trace every internal message call. A reentrancy bug hides in the order of state updates, not in the visible function names. The same is true here. The block looks clean. The fee-recipient totals look large. The missing transfer is the vulnerability.
The Data Set Is an Arbitrage Sample, Not the Whole Network
There is a second caveat hiding inside the 5:1 ratio. The Bitquery study is an arbitrage study. It counts a slice of MEV. It is not a census of every extraction strategy on Ethereum.
MEV includes arbitrage, but it also includes liquidations, sandwich attacks, NFT buy-side sniping, and censorship or ordering strategies that respond to oracle updates. Some of these strategies are easier to detect than others. Arbitrage leaves a clear trail: a searcher buys one asset in one market and sells it in another within the same block. Liquidation strategies are similarly visible through lending protocol events. Sandwich attacks are visible if you know how to cluster addresses. But not all value extraction is equally visible, and not all searchers submit bundles to Flashbots.
If the numerator of the 5:1 ratio is built from a subset of arbitrage transactions, then the ratio does not describe Ethereum's full MEV economy. It describes only the arbitrage that was visible to one data provider. That would make the number a lower bound for total MEV activity, not a precise measurement. Worse, if the denominator is total EIP-1559 burn across all network activity, the comparison mixes an incomplete sample with a complete network-wide denominator. That is like dividing a local police precinct's arrest count by the national crime rate.
The comparison across chains adds another wrinkle. The annual survey covered Ethereum, BNB Chain, Base, Arbitrum, and Polygon. It excluded Optimism and Solana. Solana's absence is especially awkward because Solana is where high-frequency trading and MEV-like value extraction have exploded. The absence may be due to data availability constraints, but the report's public summary does not explain why. A chain comparison that drops Solana is not a neutral empirical choice. It is a selection decision that changes the conclusions.
There is also a question of pricing consistency. Fixed reference prices do not move with historical exchange rates. If a data set values a February transaction using an August ETH price, or uses one reference price for an entire time series, then cross-month comparisons become distorted. CryptoSlate noted an inconsistency between fixed reference prices and historical sequence prices. That discrepancy matters because MEV dollar values are volatile even when the underlying ETH flow is stable. A transaction that generated 0.01 ETH in January and a transaction that generated 0.01 ETH in August can look like completely different dollar values depending on the pricing methodology.
None of these caveats are fatal by themselves. But they compound. Missing proposer payments, an arbitrage-only sample, excluded chains, and inconsistent price references do not make 5:1 false. They make it unverifiable. And an unverifiable ratio should not drive an Ethereum supply narrative.
The Burn Narrative Needs an Issuance Counterfactual
Now the harder problem. Even if the data were perfectly transparent, the 5:1 ratio still would not mean what many readers think it means.
EIP-1559 burns base fees. This is real. Every block destroys a portion of the ETH paid as the base fee, and that destruction reduces the ETH supply compared with a world in which those fees were paid to validators. But a burn is not automatically a supply contraction.
Ethereum also issues new ETH to validators. PoS issuance rewards attestation activity and block proposals. At the same time that base fees are being destroyed, new ETH is entering circulation. The supply effect of EIP-1559 depends on whether the burn rate exceeds the issuance rate over a given period.
This is where the popular narrative collapses. The market likes to say that Ethereum is deflationary because it burns ETH. That statement is incomplete. Ethereum is deflationary only when burn exceeds issuance. During low-activity periods, issuance can outpace burn. During high-activity periods, burn can outpace issuance. The network's supply trajectory is an equilibrium between two opposing flows, not a simple consequence of burning.
The useful metric for a holder is not a 5:1 ratio between builder fees and burned fees. The useful metric is the rate of ETH burned versus the rate of ETH newly issued within a defined period. I run this query in Dune every week. It tells me whether the supply is expanding, contracting, or holding steady. It answers the question that matters. The ratio between two unrelated fee flows does not answer that question.
The report itself appears to acknowledge this. The 5:1 headline carries a big caveat: a large inclusion payment can change the amounts that participants receive without producing an equal increase in burned ETH. In other words, a searcher paying a builder extra priority fees does not automatically cause more base fees to be destroyed. The base fee is determined by block demand. The priority fee is a payment for ordering and inclusion. These are separate mechanisms. A market that sees 5:1 as proof of a burn-driven supply squeeze is confusing priority fees with base fees.
I have seen this confusion before. During DeFi Summer in 2020, I searched through Uniswap v2 pools and found yield farms that were pure mint functions in disguise. Twenty percent annualized yields were often just inflation being paid to early depositors. The market called it value creation. The ledger called it supply dilution. EIP-1559 burn narratives can suffer from the same optical illusion. A fee can be destroyed without that destruction being large enough to offset issuance. A fee can also be destroyed while a much larger ETH payment flows to builders. The destruction is real, but it is only half the supply equation.
Another hidden issue is the relationship between transaction count and burn. Transaction counts are not gas consumption. A network can process millions of cheap transfers and burn less than a network processing a thousand complex swaps. The base fee also adjusts based on the gas used in the previous block. If demand is low, the base fee drops, and each individual transaction burns less. Looking at transaction volume as a proxy for ETH burn is a methodological error.
The report's decision to count transaction totals, if that is what drove the 5:1 headline, would explain why the ratio could be overstated. Builders can receive fees from many transactions without the network burning meaningful ETH. High transaction volume on L2s, for instance, can compress mainnet burn while builders on the mainnet still process relatively few but complex blocks. The ratio would then reflect an activity mismatch rather than an economic law.
For passive ETH holders, the burn narrative has another uncomfortable feature. Burning does not transfer cash to holders. It reduces the total available supply over time, but it does not write a check to anyone. A holder cannot claim a share of the burned base fee. The benefit of burn is only realized indirectly through the supply-demand equilibrium of the entire asset. That benefit is real, but it is not the cash distribution that the 5:1 headline implies. A protocol can burn enormous value while holders still see their position decline because issuance is larger, because market sentiment turns, or because capital rotates elsewhere.
The report is most useful when it forces this distinction. Trading activity and holding ETH are different investment frameworks. A trader cares about latency, liquidity, and fee markets. A holder cares about issuance, burn, and net supply. Combining those two frameworks into a single ratio produces a number that is intellectually convenient but analytically hollow.
Why the Contrarian Read Is Not the Obvious One
The temptation is to read the 5:1 ratio as an indictment of MEV. Look, ETH is being burned, and builders are walking away with five times more. That sounds like organized extraction. It is the kind of story that makes regulators pay attention and retail users feel like victims.
I think that read is too easy.
Arbitrage is not pure theft. When a searcher spots a DEX trading ETH at a stale price and trades against it, that searcher is moving the market closer to equilibrium. Arbitrageurs are unpaid market makers in a sense. They earn profit by correcting price distortions. Without them, decentralized exchanges would be easily arbitraged by insiders and slow-moving liquidity providers would bear larger losses. MEV has a genuine market-efficiency component.
The 5:1 ratio, if it reflects a robust arbitrage economy, is not automatically evidence of a broken network. It is evidence that Ethereum has enough liquidity, activity, and heterogeneity to create price discovery opportunities. The problem is not the existence of MEV. The problem is opacity in how MEV income is allocated. The same Flashbots ecosystem that professionalized MEV also created a centralized bottleneck. Flashbots relays have become the dominant path through which builders deliver blocks to proposers. A handful of relay operators can see the order flow. A handful of builders can dominate block construction.
That centralization has its own economic consequence. When the builder ecosystem is narrow, the competition for proposer payments can decline. A dominant builder can pass less value to validators. A relay can fail to route a block and cause proposers to miss slots. The market infrastructure that makes MEV transparent is also the market infrastructure that makes Ethereum's block production dependent on trusted intermediaries.
The New York Fed Staff Report is a signal that this problem has moved beyond crypto Twitter. Central banks do not publish research on MEV because they find sandwich attacks amusing. They publish it because block-building auctions affect how transaction orders are selected, and transaction order selection affects market integrity. A headline ratio that treats builders as the main winners could invite a regulatory response aimed at the wrong target. Regulation may come for builders when the actual bottleneck is the relay layer.
So my contrarian position is this: the 5:1 ratio is probably too high, but the underlying trend is not fake. Builders do capture a meaningful share of MEV-related fee flow. The ethical issue is not that builders earn money. The ethical issue is that the payment chain is not tracked by most public analytics. That lack of tracking allows the market to use a gross flow number as if it were a profit number. A forensic correction could lower the headline ratio without changing the underlying network.
I also want to resist the reflex to say that Bitquery is deliberately misleading the market. In my experience, data providers are rarely dishonest. They are usually constrained by schema design. If their internal architecture labels the builder fee recipient as the terminal owner, the architecture will keep telling that story even when a trace-level query would reveal outgoing payments. Fixing the story requires infrastructure work, not just a new headline. It requires tracing each builder's fee recipient through all subsequent transfers inside the same block.
The good news is that the raw data exists. Ethereum blocks are public. I can query block traces and see transfers from fee-recipient addresses to validator addresses. Independent analysts can reconstruct the second leg. What is missing is not the truth. What is missing is a broad social consensus that the second leg must be counted before anyone repeats the ratio.
What I Would Actually Watch Now
Forget the 5:1 ratio. In a sideways market, narratives calcify quickly. This one has already calcified around an incomplete ledger. The better question is where to put your attention while the market waits for directional confirmation.
The first signal is the net burn rate, not the gross burn headline. Look at ETH issued versus ETH burned over a rolling 30-day and 90-day window. If burn consistently exceeds issuance, the deflationary story has real weight. If issuance overtakes burn during quiet periods, the story does not.
The second signal is builder-to-proposer payment data. If Flashbots or a third-party research group starts publishing the total volume that builders pay to proposers, we will finally have a clean denominator for builder profit. When that number appears, expect someone to revise the 5:1 headline downward. That revision will not mean Ethereum is less active. It will mean the market finally counted the full transaction chain.
The third signal is relay diversity. Watch whether block proposal slashing events expose relay failures. Watch whether Ethereum moves toward inclusion lists or other mechanisms that reduce builder power. Those changes matter more for MEV economics than any single research ratio. The actual question is not how many dollars go to builders versus burn. The question is whether the market can observe those dollar flows at all.
I have been through enough market cycles to know that a narrative can outlive its data. The 5:1 ratio has a clean shape, but a clean shape is not a clean ledger. A receipt is not a profit. A fee recipient is not a final owner. An untraced transfer is not a finding. And a number that ignores the proposer is not a conclusion.
Follow the gas, not the narrative. The burn is visible. The builder's fee intake is visible. The payment to the validator is visible too. It is on-chain. It is traceable. It has simply not become part of the public accounting yet.
When it does, the story will change. Not because Ethereum's MEV economy became smaller, but because the market finally learned to count the second leg. That correction will be the real news of the next cycle.
