The House Edge

Appearance

7. Exclusive Order Flow ​

The contracts behind the dollar — and the reason a top-quintile validator is no longer just a better-run one. ​

Cold open ​

Users of the Banana Gun trading bot on Ethereum paid 4,466.89 ETH in fees over a recent measured window. 2,915.65 ETH reached the validators. 2,271.26 ETH stayed with one firm: a block builder called Titan.[1] The arrangement is not concealed. Banana Gun routes every user transaction to Titan, exclusively. In exchange, Titan pays validators substantially less than a non-exclusive route would have produced and keeps the difference. Titan's reported margin under the deal: 17.75%. Beaverbuild's pre-migration margin: 9%. Flashbots' historical margin: near zero. Before the contract, Titan built under 1% of Ethereum blocks. After, ~40%. By May 2026, ~52%.[2]

This chapter is about that contract. About the half-dozen comparable contracts on Solana and Ethereum that the public has never been shown. And about an empirical claim those contracts support: on most chains today, the difference between a top-quintile and bottom-quintile validator is not operational excellence. It is access.

What this chapter answers ​

Four threads. What exclusive order flow is, mechanically, against the public-mempool architecture the chains' founding documents promised. Which arrangements are documented: Banana Gun and Titan, Maestro and Beaverbuild, Helius and the named wallets, DeezNode and Vpe. What access to those arrangements is worth at the validator layer, in dollars. And how the on-chain shape compares to payment for order flow in US equities, where the EU and the SEC have spent the last eighteen months drifting in opposite directions on the same set of facts.

The setup ​

Public order flow on a permissionless chain is what the chains' founding documents promised. Any user can submit a transaction. Any searcher, validator, or RPC operator can read the public transmission paths. Block construction is allocated by the protocol's stake-weighted leader schedule. No actor has privileged sight of pending flow. The same premise animated TradFi's electronic-exchange architecture in the 1990s, when the SEC's Regulation ATS opened up alternative trading systems against the established floor brokers.

Exclusive order flow is the opposite. An infrastructure firm contracts privately with another party (a trading bot, a searcher, a wallet, a validator) to route a slice of flow on preferential terms. The contract is private. No regulator monitors it. The retail trader whose transaction it routes will not see it. Three structural shapes have emerged by 2026:

  • Direct exclusive arrangements. A block builder contracts with a specific trading firm or bot to route its transactions to that builder alone, in exchange for a margin uplift. Banana Gun / Titan is the canonical published case.
  • Pipeline relationships with rebates. An RPC provider contracts with a wallet to share MEV captured by routing transactions through the RPC's preferred path. Helius's 50/50 wallet-rebate split with Phantom, Backpack, Solflare and the other named wallets is the canonical case on Solana.
  • Validator-side fragmented surfaces. A specific validator (or pool of operators) contracts with a private-mempool operator or searcher firm for a revenue-share arrangement that re-creates the kind of pre-leader visibility the Solana Foundation enforced against in June 2024. DeezNode + Vpe from Chapter 3 is the canonical case before that enforcement.

The claim these arrangements support runs through Chapters 5 and 6 and lands here: on most chains today, the difference between a top-quintile and bottom-quintile validator is not operational excellence. It is access. The empirical decomposition is straightforward. On Solana in 2026, client-choice and infrastructure-relationship factors produce +33% to +101% per-block priority-fee revenue versus the median. Operational and timing-game optimisation factors produce ~+3%. The access-to-operational ratio is roughly an order of magnitude.

The closest TradFi parallel is payment for order flow. Same concentration shape; opposite regulatory direction. The EU's PFOF ban takes effect on 30 June 2026, six weeks after this chapter is written. The SEC withdrew its Order Competition Rule on 12 June 2025. Two regulators looking at the same structural fact, drifting opposite ways inside eighteen months. The on-chain version has no regulator at all.

The worked example ​

Banana Gun is a Telegram-based trading bot that processes hundreds of millions of dollars of Ethereum DEX swaps per month for its users. Banana Gun's commercial model: users pay a fee on each swap (a percentage of the transaction); Banana Gun captures that fee plus any MEV value its routing creates; Banana Gun then routes the bundled transaction to one specific block builder under an exclusive arrangement.

That builder is Titan. The arrangement, documented in industry coverage and confirmed by Banana Gun's own performance disclosures, routes all Banana Gun user transactions to Titan via private RPC. Titan in exchange pays Banana Gun's users — through their proposers — substantially less of the captured MEV value than a competitive routing path would have produced. The published dollar trace: of 4,466.89 ETH that Banana Gun users paid as fees and slippage across the measured window, 2,915.65 ETH reached Ethereum proposers (the validators) and approximately 2,271.26 ETH was retained, exclusively, by Titan.[1:1]

The structural consequence the chapter's cold open framed: before the arrangement, Titan built less than 1% of Ethereum blocks. After the arrangement, Titan's builder share rose to approximately 40%. By May 2026, Titan's share was approximately 52% of all Ethereum mainnet blocks (relayscan.io live snapshot, 14 May 2026). Titan's reported builder margin under the Banana Gun deal is approximately 17.75% — against Beaverbuild's pre-BuilderNet-migration margin of approximately 9% and Flashbots' near-zero historical margin under its non-exclusive model.[2:1]

The chapter's question — asked once and asked again at the verdict — is: which party in this four-actor pipeline (Banana Gun's user → Banana Gun → Titan → the proposer) is paying for the additional 17.75% of margin Titan captures? The answer the chapter develops is that the user is paying, in the form of worse execution than a competitive route would have produced; the proposer's revenue is less than it would have been under a non-exclusive arrangement; and the structural beneficiary is a single block-builder firm whose share of the network's block construction rose from less than 1% to over 50% on the strength of one contract.

The mechanics, in detail ​

Direct exclusive arrangements: the Ethereum builder-trading-firm model ​

The Banana Gun / Titan arrangement is not isolated. Academic research has identified the structural shape across the Ethereum builder market.

Wu et al.'s peer-reviewed analysis of Ethereum's builder market identified 75 distinct Exclusive Order Flow arrangements collectively accounting for approximately 71% of trading-related Ethereum builder revenue. The top five "pivotal providers" of exclusive flow included Banana Gun (paired with Titan, responsible for approximately 37.1% of estimated proposer losses from exclusive arrangements) and Maestro (paired with Beaverbuild, approximately 6% of proposer losses).[3] The Maestro-Beaverbuild pairing is the second most-cited exclusive arrangement on Ethereum; together with Banana Gun-Titan, these two contracts account for the great majority of the chapter's empirical structural argument on the Ethereum side.

Pahari and Canidio's November 2025 measurement of Ethereum exclusivity — using Agnostic Relay data from December 2024 — produced a more aggregate stat: exclusive transactions account for 77.2% to 84% of total fees paid in winning Ethereum blocks. The pair found that only approximately 7%–8.4% of on-chain exclusive transaction value comes from senders routing exclusively to one builder; most exclusivity is dynamic rather than contractually locked. But the aggregate fee impact is concentrated in transactions that are not available through public order flow — meaning the public mempool, in 2026, contains a structural minority of the network's fee-paying value.[4]

The arrangements are not formally disclosed in any regulatory filing. They exist as private commercial contracts between named trading firms and named builders. The Wu et al. paper assembled its findings through on-chain analysis — observing which transactions consistently appeared in which builder's blocks, then inferring the underlying contracts — because no public disclosure regime requires the contracts themselves to be filed.

Pipeline relationships: the Helius / wallet-rebate model ​

On the Solana side, the dominant exclusive-flow pattern operates at the RPC and wallet layer rather than at the builder layer. Helius's 50/50 wallet-rebate split — already developed in Chapter 6 — is the canonical pipeline relationship. The mechanism: a transaction routed through Helius's RPC is exposed to a private auction with pre-approved KYC'd searchers (post-trade backruns only, paid in SOL in the same block); the resulting MEV value is split 50/50 between Helius and the wallet that submitted the transaction.[5]

The named wallet customers per Helius's published Wallets product page, as of 14 May 2026, are eight: Phantom, Ledger, Bitgo, Exodus, Backpack, Squads, Trust, and Solflare. (Bitwise is a separately documented Helius partnership — the exclusive Solana ETF staking provider — and is not on the wallet-customer list itself.) The structural fact about the relationship: most of the wallets the Solana retail trader has ever heard of share a single RPC backend, and that backend's commercial arrangement with the wallets is the same 50/50 rebate structure across all of them. Helius is structurally the Solana-side analogue of a US wholesaler-of-retail-flow — except that the rebate flows back toward the user rather than directly to the broker, and the commercial terms are not regulated by anyone.

The aggregate dollar volume passing through the Helius rebate pipeline in 2026 is not publicly disclosed, and the downstream split between Helius and its wallet customers — and whether the wallets pass the rebate through to their users or retain it as platform revenue — is similarly opaque. The chapter notes the opacity as a chapter-level point. Phantom's 2025 revenue of approximately $79.1 million tracks the wallet's transaction-fee revenue weekly, confirming fees as the primary income source; the firm has not separately disclosed what fraction of that revenue is rebate-derived.[6] Backpack and Solflare have published no comparable revenue figures.

The structural contrast with the Ethereum builder-direct case: Banana Gun / Titan is a contract between two firms that the public can analyze through on-chain data. Helius / Phantom is a contract between an RPC provider and a wallet whose commercial terms can be partially inferred from product documentation but whose dollar flows are not visible on-chain because the value capture happens at the RPC layer rather than at the block-construction layer.

Validator-side fragmented surfaces: the persistent grey market ​

The third structural shape is the post-2024 grey-market private mempool surface on Solana. Chapter 3 developed the case in detail: in June 2024, the Solana Foundation removed more than 30 validators from its delegation program for participating in private mempools that enabled sandwich attacks. The named pre-shutdown operator, DeezNode (validator address HM5H6…jdMRA), ran the Vpe sandwich program — approximately 1.55 million sandwich transactions over 30 days in December 2024 to January 2025, profiting 65,880 SOL (approximately $13.43 million at the time), against approximately 811,604 SOL of delegated stake (~$168.5 million). The arrangement reportedly offered a 50% revenue share to opting-in validators.[7]

Tim Garcia, the Solana Foundation's validator-relations lead, framed the June 2024 enforcement as "ongoing." No public list of additional removals between June 2024 and May 2026 has been published. The chapter treats this as a chapter-level observation: the Foundation's enforcement is asserted but not publicly itemised after the initial action. The grey-market surface that gave rise to DeezNode in 2024 has not, by the foundation's own framing, been eliminated; it has simply been driven into less-public arrangements.

The structural argument the chapter has been building lands here. The Foundation's June 2024 enforcement removed visible operators but did not eliminate the demand for the visibility surface. The market for pre-leader transaction sight — which Jito Labs operated openly as a paid Block Engine product through March 2024 — has not disappeared. It has fragmented into RPC-layer pipeline relationships (Helius and the wallet customers), into BAM-layer confidential surfaces (Chapter 3's TEE-encrypted block construction), and into the less-public arrangements between specific validators and specific searcher firms that the foundation's enforcement is, in Garcia's framing, still pursuing.

The validator-quintile decomposition: the empirical argument ​

The chapter's load-bearing structural claim, framed by the book's outline: the difference between a top-quintile and bottom-quintile validator on most chains is no longer operational excellence — it is access to exclusive flow. The empirical evidence supporting this claim, assembled from three sourced 2025–2026 measurements:

The "access" signal: client-choice and infrastructure-relationship factors. Syndica's March 2026 deep dive measured per-block priority-fee captures for Solana validators running different validator-client + scheduler configurations against the network median. Frankendancer Harmonic Performance (validators routing through Harmonic's MREV strategy) captured approximately +101% priority fees per block; Frankendancer Harmonic Balanced approximately +39%; Agave Harmonic approximately +36%. Each of these is a function of the validator's choice of client software and its participation in the Harmonic block-building marketplace — both pure access decisions, available to any validator with the operational sophistication to make them. Block-time-adjusted figures are slightly lower but remain materially above network median: Agave Harmonic +33%, Frankendancer Harmonic Balanced +32%, Performance +28%.[8]

The "operational" signal: hardware, latency, and timing-game optimisation. Chorus One's August 2025 empirical research on Solana timing games measured the combined revenue gain from block-production timing manipulation and scheduler optimisation. The total: approximately +3.0% in total rewards (about 27 basis points annualised). Timing games alone produced approximately +1.19% (~8 bps); scheduler optimisation alone produced approximately +1.4% (~12 bps). These are the gains available to a validator that exhausts every operational optimisation publicly known.[9]

The comparison: access produces approximately +33% to +101%; operational excellence produces approximately +3%. The ratio is approximately an order of magnitude. The chapter's central claim is empirically correct, and the order-of-magnitude gap is the strongest single piece of analytical work in this section of the book.

The dispersion is reinforced by Placeholder VC's September 2025 measurement of the Solana validator profit distribution: the Gini coefficient for validator profits is approximately 0.93 — extreme inequality. (For comparison, the most unequal US states by household income have Gini coefficients around 0.5; the most unequal countries globally are around 0.6.) The top-100 validators receive substantial inflation-based rewards; the bottom-100 rely almost entirely on transaction fees and MEV revenue when selected as slot leader. Low-stake validators are forced to pass close to all value back to delegators.[10]

The institutional case study: Figment's Q4 2025 Solana Validator Report documented a Staking Reward Rate of 6.44% against the network average of 4.70% — a 31% outperformance. Jito MEV contributed approximately 0.2% of total staking rewards in Q4 2025, projected to reach 20–25% of validator rewards in high-activity windows as Solana's annual −15% inflation schedule continues to compress the issuance share.[11] Figment's outperformance is not, in the report's framing, primarily a function of operational excellence (the firm's uptime and skip rate are excellent but not unique). It is a function of access: Figment was a launch operator for Jito BAM, runs Frankendancer, migrated to Rakurai on 2 March 2026, and maintains active relationships with the major Solana infrastructure firms.

Top-quintile validatorMedian validatorBottom-quintile validator
Client choiceFrankendancer / Rakurai / JitoBAM with Harmonic strategyAgave Jito (stock)Agave Jito (stock) or older
Builder/relay relationshipsActive (Jito BAM Node + Harmonic preferences set)Default Block Engine routingDefault Block Engine routing
Approximate per-block revenue vs median+33% to +101% (Harmonic Performance high end)BaselineBaseline or below
Marginal gain from timing games + scheduler optimisation+~3% additional+~3% additional+~3% additional
Implied attribution of top-vs-bottom gapAccess (client choice + builder relationships) dominates; operational factors secondaryn/an/a
Gini contribution to network-level inequalityTop contributor (extreme upper tail)MedianBottom contributor; relies on inflation rewards

The empirical conclusion: the gap between a Solana top-quintile and bottom-quintile validator is, in 2026, approximately one order of magnitude larger on the access dimension than on the operational dimension. The OUTLINE's central claim is correct. The chapter has sourced it.

The TradFi parallel: payment for order flow in 2026 ​

The structural similarity to the US equity wholesale market is sharp. Chapter 1 cited the SEC's January 2025 DERA working paper on PFOF, which assembled the wholesaler-concentration data: Citadel Securities at approximately 41% of US retail equity wholesale flow, Virtu Financial at approximately 26%, G1 Execution Services at approximately 16% — three firms accounting for more than 80% of the market.[12] The Ethereum builder-share figures from Chapter 6: Titan ~52%, BuilderNet ~24%, Quasar ~15% — three firms accounting for approximately 91%.

The shapes are similar. The regulatory regimes are not.

In TradFi, the practice is regulated at multiple levels. SEC Rule 605 requires market centers to publish monthly execution-quality reports; Rule 606 requires broker-dealers to disclose order-routing arrangements quarterly; the SEC's National Best Bid and Offer (NBBO) framework establishes baseline best-execution obligations. In 2026, the regulatory direction in the US and the EU has bifurcated sharply.

The European Union definitively bans payment for order flow effective 30 June 2026 — six weeks after the publication of this chapter. The ban was agreed under MiFIR amendments in 2023 and entered into force in March 2024 with a transitional period during which member states could exempt domestic firms. Only Germany exercised the exemption (notified to ESMA in March 2024), primarily to give German neobrokers like Trade Republic a two-year runway. After 30 June 2026, PFOF is prohibited bloc-wide across all EU member states.[13]

The United States has moved in the opposite direction. On 12 June 2025, the SEC under chair Paul Atkins formally withdrew fourteen Gensler-era rule proposals, including the Order Competition Rule (which would have required wholesalers to expose retail "segmented orders" to 100–300 millisecond auctions on open venues) and Regulation Best Execution. With the withdrawals, "the foundational business model of bilateral agreements between brokers and wholesalers can continue unimpeded." No new PFOF-restricting rulemaking is pending under the current SEC.[14]

Two financial-system regulators looking at the same structural fact — concentrated wholesale flow with private commercial arrangements between brokers and wholesalers — have, in 2025–2026, taken opposite positions. The European Union has decided this structure is bad enough that it must be banned. The United States has decided this structure is acceptable enough that proposed restrictions on it should be withdrawn.

The on-chain version of the same structure operates under neither regime. There is no analogue to SEC Rule 605 disclosure for Ethereum builder-direct relationships. There is no equivalent of the EU PFOF ban for Solana RPC-layer pipeline rebates. There is no regulator to which Helius, Titan, or Banana Gun discloses the commercial terms of its arrangements. The structural similarities to TradFi PFOF — concentration, opacity, value flowing back to the user but at terms the user did not negotiate — exist; the regulatory infrastructure that exists for TradFi PFOF (even in its most permissive 2025-Atkins form) does not.

How this plays out on each chain — and Hyperliquid as the structural counterpoint ​

On Solana, exclusive flow operates as a mix of pipeline relationships (Helius and the wallet customers; 50/50 rebate splits), validator-side fragmented surfaces (the post-June-2024 grey-market private mempools whose status the Foundation describes as "ongoing enforcement" without publishing a list), and the Block Engine / BAM stack operated by Jito Labs. The chain has no equivalent of Ethereum's builder-direct contract — partly because Solana's block-construction layer is more vertically integrated (the validator is the slot leader is the block constructor) and partly because the chain's foundation has signalled it will enforce against the most extractive surfaces.

On Hyperliquid, the chain's architecture eliminates the traditional exclusive-flow surface. The chain's matching engine sits inside consensus; there is no separate block-construction layer for a third party to contract exclusively with. Bitwise's April 2026 amendment to its proposed Hyperliquid ETF S-1 names Wintermute, Flowdesk, Nonco, and FalconX as approved trading counterparties — but the filing explicitly states there is no obligation on the sponsor to route every transaction through any specific firm. This is approval-not-exclusivity: the firms have passed Bitwise's diligence to handle counterparty trading on the ETF's behalf, but the ETF can transact with any of them.[15] HIP-3 builder codes — Hyperliquid's permissioned-but-non-exclusive primitive for deploying perpetual markets — operate at a different layer (market deployment, not transaction routing) and are not structurally exclusive arrangements. Chapter 9 develops both in full.

On Ethereum and its L2s, exclusive flow is the dominant structural pattern at the builder layer. The 75 EOF arrangements that Wu et al. identified, the 77.2%–84% of fee-paying transactions that Pahari and Canidio measured, the Titan ~52% / BuilderNet ~24% / Quasar ~15% builder concentration — together they document Ethereum as the chain where exclusive flow has reached its most mature form. BuilderNet's "open-source refund rule" is the explicit structural counter-attempt: a multi-operator builder protocol that has committed to non-exclusive flow. The result, two and a half years after its November 2024 launch, is that BuilderNet's growth has come at the expense of other non-exclusive builders (Beaverbuild migrated its standalone operation into BuilderNet in May 2025), not at the expense of Titan. The structural lesson: exclusive flow is competitively durable; a non-exclusive multi-operator protocol can capture share from other non-exclusive operators but does not, in the data this chapter has assembled, appear to dent the exclusive-flow incumbent.

Who wins, who loses, why ​

Winners. Titan and Gattaca: ~52% of Ethereum's block construction on the strength of one exclusive contract with Banana Gun, reportedly at 17.75% margin. Maestro and Beaverbuild: the structurally similar smaller-scale case, ~6% of estimated proposer losses per Wu et al. Helius and its named wallet customers, sharing the 50/50 rebate split that makes the Solana pipeline relationship work. Jito Labs, whose Block Engine and BAM stack operate as the same kind of exclusive surface on Solana, with JIP-24 routing the take rate to the DAO instead of the firm (Chapter 6). Banana Gun and Maestro themselves, who built bot businesses on top of those arrangements. SCP and Barter at the solver layer: the duopoly hiding inside UniswapX and CoW Swap, with SCP and Wintermute taking >90% of UniswapX volume and Barter targeting >50% of CoW share. And Lido's curated ~36-operator institutional set, sitting on privileged access to the protocol's stake pool.

Losers. Bottom-quintile validators are the primary loser. The access gap to top-quintile is roughly an order of magnitude larger than the operational gap, which means a validator running every publicly known operational optimisation still earns materially less than one with the right client choice and the right infrastructure relationships. Banana Gun's users are the named loser in the cold open: the 2,271.26 ETH Titan retained came out of the execution price they got. Retail traders generally, every layer of the stack pulling a cut. And the public-goods narrative the chains' founding documents committed to (anyone can route a transaction through a competitive market) no longer describes how the market operates for most fee-paying transactions on Ethereum.

Is any of this bad? Take the clinical view. The structural shape (direct exclusives, pipeline rebates, fragmented validator-side surfaces) is the book's third documented case of permissionless chains producing institutional concentration their founding documents did not anticipate. Chapter 2 was the first: prop-AMM displacement of public LPs. Chapter 6 was the second: the infrastructure stack. Chapter 7 is the third: exclusive contracts between named firms that the public cannot see, in arrangements no regulator monitors. The book's outline called this "the most important undiscussed dynamic in validator economics." The data supports that framing. The access-to-operational ratio for Solana validator revenue is roughly 10×. The Gini coefficient for Solana validator profits is ~0.93. The 75 exclusive arrangements Wu et al. identified account for ~71% of trading-related Ethereum builder revenue. These are not anomalies. They are the structural shape the network's institutional layer has taken since 2023.

No individual exclusive arrangement here is illegitimate. Banana Gun and Titan have a legitimate commercial agreement. So do Helius and Phantom. The Solana Foundation's enforcement against DeezNode-style validator arrangements is the foundation operating inside its delegation-program authority. The point is the aggregate. The data cited in this chapter says the aggregate effect of these arrangements has produced a market structure more concentrated than US equity market structure under the most permissive interpretations of the post-Atkins SEC, and unambiguously more concentrated than the EU's market structure will be after 30 June 2026. The structure is not unregulated because no one has considered regulating it. It is unregulated because no regulator has jurisdiction.

What changes when… ​

The actor chapters end here. Chapters 8 through 10 (the chain-specific deep dives) pick up the contracts, pipelines, and grey-market surfaces this chapter has documented and place them against the specific architectures each chain has built. Solana's combination of Gulf Stream + Jito Block Engine + BAM + Harmonic + the Foundation's enforcement, read as a system. Hyperliquid's HyperBFT consensus + HLP + HIP-3 + the 21-validator permissioned set, read as a competitive equilibrium. Ethereum's MEV-Boost + the slipping ePBS roadmap + Titan's 52% + BuilderNet's 24% + the L2 sequencer landscape, read together. Those are the questions of Part III.

Footnotes and sources ​

  1. Coverage of the Banana Gun / Titan exclusive arrangement: Techub News / PANews analysis, How the Banana Gun siphoned millions of dollars from users and Ethereum validators, via Gate.io aggregation, https://www.gate.io/learn/articles/how-the-banana-gun-siphoned-millions-of-dollars-from-users-and-ethereum-validators/3815. The 4,466.89 ETH / 2,915.65 ETH / 2,271.26 ETH dollar trace is from the measured window documented in that analysis. Accessed 2026-05-14. ↩︎ ↩︎

  2. Observers, How Two Block Builders Monopolized Ethereum Block Production, https://www.observers.com/how-two-block-builders-monopolized-ethereum-block-production/, citing Titan disclosures: ~17.75% margin under the Banana Gun arrangement versus Beaverbuild's pre-migration ~9% and Flashbots' historically near-zero. The <1% → ~40% builder share trajectory after the Banana Gun deal is documented in the same source; the May 2026 ~52.16% builder share is from relayscan.io (24-hour snapshot ending 14 May 2026). Already cited in Chapters 5 and 6. Accessed 2026-05-14. ↩︎ ↩︎

  3. Wu, Brünjes, Wattenhofer et al., Decentralization of Ethereum's Builder Market, arXiv:2405.01329 (continuing analysis through 2025), https://arxiv.org/html/2405.01329v4. 75 Exclusive Order Flow arrangements accounting for approximately 71% of trading-related Ethereum builder revenue. Top five pivotal providers include Banana Gun (paired with Titan, ~37.1% of estimated proposer losses) and Maestro (paired with Beaverbuild, ~6% of estimated proposer losses). Accessed 2026-05-14. ↩︎

  4. Pahari & Canidio, How Exclusive are Ethereum Transactions? Evidence from non-winning blocks, arXiv:2509.16052, revised 14 November 2025, https://arxiv.org/abs/2509.16052. Data window: December 2024. Exclusive transactions account for 77.2%–84% of total fees paid in winning Ethereum blocks; only approximately 7%–8.4% of on-chain exclusive transaction value comes from senders routing exclusively to one builder (most exclusivity is dynamic, not contractually locked). Titan held 46.5% of blocks in the sample; Rsync 15.6%; Flashbots 12.8%; Beaverbuild 9.4%. Accessed 2026-05-14. ↩︎

  5. Helius, Backrun Rebates Documentation, https://www.helius.dev/docs/sending-transactions/backrun-rebates; Helius, Wallets Use Case, https://www.helius.dev/use-case/wallets. 50/50 split between Helius and the integrator (wallet); KYC'd searcher pool; backrun-only auctions paid in SOL in the same block. Named wallet customers per the Helius Wallets product page: Phantom, Ledger, Bitgo, Exodus, Backpack, Squads, Trust, Solflare. Helius's exclusive Solana ETF staking partnership with Bitwise is a separately documented relationship (see Helius, Bitwise Solana ETF, https://www.helius.dev/blog/bitwise-solana-etf) and is not on the wallet-customer list itself. Already cited in Chapter 6. Accessed 2026-05-14. ↩︎

  6. SQ Magazine, Phantom Wallet Statistics 2026, https://sqmagazine.co.uk/phantom-wallet-statistics/. Phantom's reported 2025 annual revenue: approximately $79.1 million; peak weekly $44.14 million. The firm has not separately disclosed what fraction of that revenue is rebate-derived. Phantom's PSOL liquid-staking product shares MEV tips and priority fees with stakers. Accessed 2026-05-14. ↩︎

  7. Helius Research, Solana MEV Report, https://www.helius.dev/blog/solana-mev-report; CoinDesk, Solana Heavyweights Wage War Against Private Mempool Operators, 10 June 2024, https://www.coindesk.com/business/2024/06/10/solana-heavyweights-wage-war-against-private-mempool-operators; Anarcaze / Medium, Spotting Shady Validators on Solana, https://anarcaze.medium.com/spotting-shady-validators-on-solana-a-friendly-guide-f39ef8b32a00. DeezNode validator address HM5H6…jdMRA; 1.55M sandwich transactions over 30 days (Dec 7 2024 – Jan 5 2025) for ~65,880 SOL profit (~$13.43M); 811,604 SOL delegated stake (~$168.5M); reported 50% revenue-share offer to opting-in validators. Foundation's June 2024 removal of >30 validators; Tim Garcia's "enforcement is ongoing" framing. No public list of removals after June 2024. Already cited in Chapter 3. Accessed 2026-05-14. ↩︎

  8. Syndica, Deep Dive: Solana Onchain Activity, March 2026, https://blog.syndica.io/deep-dive-solana-onchain-activity/. Per-block priority-fee captures versus the network median: Frankendancer Harmonic Performance +101%, Frankendancer Harmonic Balanced +39%, Agave Harmonic +36%. Block-time-adjusted: Agave Harmonic +33%, Frankendancer Harmonic Balanced +32%, Frankendancer Harmonic Performance +28%. Already cited in Chapters 5 and 6. Accessed 2026-05-14. ↩︎

  9. Chorus One, Timing Games on Solana: Validator Incentives, Network Impacts, and Agave's Hidden Inefficiencies, 13 August 2025, https://chorus.one/reports-research/timing-games-on-solana-validator-incentives-network-impacts-and-agaves-hidden-inefficiencies. Combined timing-games and scheduler optimisation: approximately +3.0% in total rewards (about 27 basis points annualised). Timing games alone approximately +1.19% (~8 bps); scheduler optimisation alone approximately +1.4% (~12 bps). Already cited in Chapter 5. Accessed 2026-05-14. ↩︎

  10. Placeholder VC (and Volt Capital reprint), Leveling the Stakes on Solana, 15 September 2025, https://www.placeholder.vc/blog/2025/9/15/leveling-the-stakes-on-solana. Solana validator profit Gini coefficient approximately 0.93 — extreme inequality. The top-100 validators receive substantial inflation-based rewards; the bottom-100 rely almost entirely on transaction fees and MEV revenue when selected as slot leader. Accessed 2026-05-14. ↩︎

  11. Figment, Q4 2025 Solana Validator Report, https://www.figment.io/insights/figments-q4-2025-solana-validator-report/. Figment SRR 6.44% vs network average 4.70% (31% outperformance). Jito MEV ~0.2% of total staking rewards in Q4 2025; projected to reach 20–25% of validator rewards in high-activity windows as the −15% annual inflation schedule continues to compress the issuance share. Accessed 2026-05-14. ↩︎

  12. Boulton, Shohfi, and Walz, How Does Payment for Order Flow Influence Markets? Evidence from Robinhood Crypto Token Introductions, U.S. Securities and Exchange Commission Division of Economic and Risk Analysis Working Paper, January 2025, https://www.sec.gov/files/dera_wp_payment-order-flow-2501.pdf. Citadel Securities ~41%, Virtu Financial ~26%, G1 Execution Services ~16% of US retail equity wholesale flow (combined >80%). Already cited in Chapter 1. Accessed 2026-05-14. ↩︎

  13. ESMA, List of EU Member States using the temporary exemption to the payment-for-order-flow (PFOF) prohibition under MiFIR, https://www.esma.europa.eu/document/list-eu-member-states-using-temporary-exemption-payment-order-flow-pfof-prohibition-under; Hogan Lovells, EU MiFIR Amendments Prohibiting Payment for Order Flow, https://www.hoganlovells.com/en/publications/eu-mifir-amendments-prohibiting-payment-for-order-flow-pfof-entered-into-force-on-28-march-2024. PFOF ban entered into force 28 March 2024 with member-state opt-out until 30 June 2026; Germany notified the only exemption (March 2024). After 30 June 2026, PFOF prohibited bloc-wide. Accessed 2026-05-14. ↩︎

  14. U.S. Securities and Exchange Commission, Notice of Withdrawal of Proposed Regulatory Actions, 12 June 2025, https://www.sec.gov/files/rules/final/2025/33-11377.pdf; Proskauer, SEC Withdraws Fourteen Rule Proposals, https://www.proskauer.com/alert/sec-withdraws-fourteen-rule-proposals; Stinson, SEC Withdraws Proposed Rules Affecting Investment Advisers, Funds, and Broker-Dealers, https://www.stinson.com/newsroom-publications-sec-withdraws-proposed-rules-affecting-investment-advisers-funds-and-broker-dealers. 14 Gensler-era proposed rules formally withdrawn 12 June 2025, including the Order Competition Rule (which would have required wholesalers to expose retail "segmented orders" to 100–300ms auctions on open venues) and Regulation Best Execution. Accessed 2026-05-14. ↩︎

  15. Crowdfund Insider, Bitwise files second amendment to Hyperliquid ETF; names Flowdesk and Wintermute as trading counterparties, April 2026, https://www.crowdfundinsider.com/2026/04/272856-bitwise-submits-latest-amendment-to-hyperliquid-etf-names-flowdesk-and-wintermute-as-trading-counterparties/. Approved trading counterparties: FalconX, Flowdesk, Nonco, Wintermute. The S-1 filing explicitly states there is no obligation on the sponsor to route every transaction through any specific firm — this is approval-not-exclusivity. Already cited in Chapter 5. Accessed 2026-05-14. ↩︎