The House Edge

Appearance

Chapter 7 — Outline (Phase 2) ​

Status: OUTLINE (Phase 2 — written by agent, awaiting Nick's review) Date: 2026-05-14 Working chapter: 07 — Exclusive Order FlowCarries forward from: RESEARCH.md

Working assumptions (carried from Phase 1) ​

Defaults from RESEARCH.md §6, treated as binding for Phase 3 unless Nick overrides.

  1. Word budget: ~7,200 words including footnotes — comparable to Ch 6. The chapter has rich primary-source material (two academic papers, the Banana Gun dollar trace, Syndica + Chorus One decomposition, Lido data, PFOF 2026 regulatory bifurcation).
  2. Worked example: Candidate A (Titan + Banana Gun dollar trace) as the named anchor in §4, with Candidate B's validator-quintile decomposition as a substantial §5d subsection.
  3. Maestro paired with Beaverbuild, not Titan — per Wu et al. The chapter does not name additional Titan exclusives beyond Banana Gun.
  4. Bitwise "approved trading counterparties" framed as approval-not-exclusivity — the structural counterpoint to Ethereum's exclusive flow.
  5. PFOF as substantial subsection — the closing mechanics subsection (§5e) before the chain-comparison. The 2026 regulatory bifurcation (EU ban 30 June 2026 / SEC withdrawal June 2025) is the chapter's strongest single regulatory anchor.
  6. HIP-3 as paragraph cameo in §6 chain-comparison; full treatment in Ch 9.
  7. Solana Foundation enforcement framed as committed-but-not-publicly-itemised post-June 2024. No specific 2025–2026 enforcement events asserted.
  8. Lido as paragraph cameo — stake-side exclusivity is structurally different from builder-flow EOF; the chapter notes the distinction but doesn't develop Lido in depth.

Title and subtitle ​

Chapter 7 — Exclusive Order FlowThe contracts behind the dollar — and the reason a top-quintile validator is no longer just a better-run one.

Cold open ​

Of the 4,466.89 ETH that users of the Banana Gun trading bot paid to settle their swaps over a recent measured window on Ethereum, approximately 2,915.65 ETH reached the network's validators — the actors the Ethereum white paper assigned the right to capture transaction fees. The remaining 2,271.26 ETH was retained, exclusively, by a single block-building firm called Titan. The arrangement is not concealed: Banana Gun routes its users' transactions to Titan and Titan alone; Titan in exchange pays the validators substantially less than a non-exclusive route would have produced. Titan's reported margin under the deal is approximately 17.75%, against Beaverbuild's pre-migration approximately 9% margin and Flashbots' near-zero historical margin under its non-exclusive model. Before the arrangement, Titan built less than 1% of Ethereum blocks. After the arrangement, Titan built approximately 40% of Ethereum blocks — and by May 2026, approximately 52%. This chapter is about the contract between Banana Gun and Titan, the comparable contracts that have emerged across both Solana and Ethereum, and the empirical claim those contracts support: that the difference between a top-quintile and bottom-quintile validator on most chains is no longer operational excellence — it is access.

(Why this open: a specific dated dollar trace from sourced reporting; one named firm pair; the structural argument framed in the closing sentence; the +17.75% margin and 1%-to-52% builder-share trajectory are the most pointed published numbers the chapter has.)

What this chapter answers ​

  • What is exclusive order flow, and how is it structurally different from public order flow?
  • Which named exclusive arrangements are publicly documented in 2026 — on Ethereum, on Solana, on Hyperliquid?
  • How much does access to exclusive flow change validator economics, in measurable terms?
  • What is the structural relationship between exclusive flow on-chain and payment-for-order-flow in US equity markets — and why has the regulatory landscape diverged so sharply between the EU and the US in 2025–2026?

Section list ​

  1. Cold open — the Banana Gun / Titan dollar trace.

  2. What this chapter answers — the four questions above.

  3. The setup (≈500 words). What exclusive order flow is, in plain terms. Three structural shapes: direct contracts (a searcher or trading firm with a specific builder); pipeline relationships (an RPC provider with a wallet, with rebates flowing back); validator-side fragmented surfaces (the grey-market private mempool patterns that survived 2024 enforcement). Plant of the chapter's load-bearing claim: top-quintile vs bottom-quintile validator difference = access, not operational excellence. The PFOF parallel introduced.

  4. The worked example (≈400 words). The Banana Gun + Titan dollar trace developed in full: 4,466.89 ETH paid; 2,915.65 ETH to proposers; 2,271.26 ETH retained by Titan; ~17.75% margin; <1% → ~40% → ~52% builder share. The chapter follows the dollar through the contract structure.

  5. The mechanics, in detail (≈4,000 words; five H4 subsections — the chapter spec template's hard limit of four is exceeded by one for the PFOF closing because the cross-disciplinary anchor is load-bearing):

    • Direct exclusive arrangements: the Ethereum builder-trading-firm model (≈900 words). Banana Gun / Titan developed in full. Maestro / Beaverbuild as the second named pair (per Wu et al.). The academic anchor: Wu et al.'s identification of 75 EOF arrangements accounting for ~71% of trading-related Ethereum builder revenue; Pahari & Canidio's measurement that 77.2%–84% of total fees in winning Ethereum blocks come from exclusive transactions. The argument: exclusive flow is not an edge case in Ethereum's market structure — it is the structural majority.

    • Pipeline relationships: the Helius / wallet-rebate model (≈800 words). Helius's 50/50 wallet-rebate split, the wallet customers (Phantom, Backpack, Solflare, Bitwise, Coinbase), the opacity of aggregate flow dollars. The Phantom $79.1M 2025 revenue figure as indirect anchor. The structural argument: pipeline relationships are contractual in the same way builder-direct exclusives are, but their commercial terms are opaque to outside observers in ways the Titan / Banana Gun arrangement is not. The downstream split between Helius and wallets, and the wallet-side decision whether to pass the rebate through to users, is not publicly disclosed.

    • Validator-side fragmented surfaces: the persistent grey market (≈600 words). The DeezNode + Vpe case from Ch 3 returns. The Solana Foundation's June 2024 enforcement against ~30 validators; the committed-but-not-publicly-itemised nature of subsequent enforcement. The argument the chapter has been building: the Foundation's enforcement removed visible operators but did not eliminate demand for the surface; the visibility-as-product market simply moved to RPC providers and wallet pipelines.

    • The validator-quintile decomposition: the empirical argument (≈900 words). The chapter's load-bearing structural claim, with sourced numbers. Top-line: client-choice gap (access) ≈ +33% to +101% per-block priority fees (Syndica March 2026); timing-and-scheduler optimisation gap (operational) ≈ +3% total rewards (Chorus One August 2025). Access dwarfs operational by an order of magnitude. The Solana validator profit Gini coefficient of 0.93 anchors the inequality scale (Placeholder VC September 2025). Figment's 6.44% SRR vs network average 4.70% (31% outperformance) as the institutional case study. The conclusion the chapter explicitly defends: the OUTLINE.md's central claim is correct and the chapter has sourced it. The validator-quintile decomposition is the chapter's strongest single piece of analytical work. The matrix table (D2) consolidates the comparison.

    • The TradFi parallel: payment for order flow in 2026 (≈800 words). The structural similarity: Ethereum builder concentration (Titan + BuilderNet + Quasar ≈ 91%) vs US wholesaler concentration (Citadel + Virtu + G1 ≈ 80%). The same shape, very different regulatory regimes. The 2026 regulatory bifurcation: the EU's PFOF ban takes effect 30 June 2026 (six weeks after the chapter's writing date), while the SEC withdrew 14 PFOF-restricting rule proposals on 12 June 2025, including the Order Competition Rule. The on-chain version has no analogue to either regulator. The conclusion: on-chain exclusive flow operates with a regulatory vacuum that even the most permissive TradFi regime (the post-Atkins US) does not match.

  6. How this plays out on each chain — and Hyperliquid as the structural counterpoint (≈400 words). Solana's mix (Helius pipeline + DeezNode-style grey market + the Foundation's selective enforcement); Ethereum's strong-form (Banana Gun / Titan direct exclusives + the 77.2%–84% of fee-paying transactions being exclusive); Hyperliquid's architectural absence of the surface (HIP-3 builder codes as a permissioned-but-non-exclusive primitive; Bitwise's "approved trading counterparties" framing as approval-not-exclusivity).

  7. Who wins, who loses, why (≈500 words). Adversarial verdict in the Ch 3 / Ch 4 / Ch 6 register.

    • Winners: Titan and Gattaca (52.16% Ethereum builder share + 17.75% margin under exclusive); Beaverbuild (the Maestro pairing); Helius (the wallet-pipeline operator); Phantom and the named wallet customers (downstream rebate capture); Jito Labs (Block Engine + BAM as a structurally similar exclusive surface on Solana); Banana Gun, Maestro, and the trading-firm/bot counterparties whose users pay the bill; SCP and Barter at the solver layer (the duopoly disguised as competition).
    • Losers: top-quintile vs bottom-quintile validators whose access gap is ~10× the operational gap; Banana Gun's own users (the 2,271.26 ETH of retained margin came out of their effective execution price); retail traders generally (every layer of the stack takes a cut); the public-goods narrative of "anyone can route a transaction through a competitive market."
    • Is this bad?: clinical. The structural shape — direct exclusives, pipeline rebates, fragmented validator-side surfaces — is the chapter's third documented case of permissionless chains producing institutional concentration that the chains' founding documents did not anticipate. The OUTLINE.md framing — that this is "the most important undiscussed dynamic in validator economics" — is supported by the data: the access-vs-operational ratio is roughly 10× and the dispersion Gini is 0.93 on Solana. The chapter does not assert that any individual arrangement is illegitimate; it asserts that the aggregate effect has produced a structural surface the trader cannot see and no regulator monitors.

  8. What changes when… (one paragraph). The transition to Part III. What changes when these exclusive arrangements, pipeline rebates, and fragmented surfaces have to coexist with the specific architectures each chain has built? When Solana's Gulf Stream + Jito Block Engine, Hyperliquid's HyperBFT consensus, and Ethereum's MEV-Boost are not abstract designs but the rails through which these contracts actually operate? That is Part III — Chapters 8 through 10.

  9. Footnotes and sources — numbered with URLs and access dates. Approximately 22–25 footnotes.

The worked example and where it threads through ​

Banana Gun + Titan as the named anchor; the validator-quintile decomposition as the empirical anchor.

SectionBeat
§3 (Setup)Plant: "we'll follow the Banana Gun / Titan dollar trace, then prove the validator-quintile claim."
§4 (Worked example)The full Banana Gun / Titan dollar trace: 4,466.89 ETH paid, 2,271.26 ETH retained by Titan exclusively, 17.75% margin, <1% → ~52% builder share trajectory.
§5a (Direct exclusives)Titan / Banana Gun developed in mechanism; Maestro / Beaverbuild as second named pair; Wu et al. + Pahari/Canidio academic anchors.
§5b (Pipeline relationships)Helius / wallet pipeline; Phantom $79.1M 2025 revenue as indirect anchor.
§5c (Validator-side fragmented surfaces)DeezNode + Vpe (returning from Ch 3); the Foundation's June 2024 enforcement and the post-enforcement opacity.
§5d (Validator-quintile decomposition)The chapter's load-bearing empirical argument: access (+33% to +101%) vs operational (+3%); Gini 0.93; Figment 31% outperformance.
§5e (PFOF parallel)Citadel + Virtu + G1 = >80% vs Titan + BuilderNet + Quasar = ~91%; the 2026 regulatory bifurcation.
§6 (Chain comparison)Solana's mix; Ethereum's strong-form; Hyperliquid's structural absence.
§7 (Verdict)Banana Gun's users as named losers (the 2,271.26 ETH came out of their pockets).

Diagrams needed ​

Two diagrams.

  1. D1 — The three exclusive-flow patterns (Mermaid flowchart with three lanes).

    • Lane 1 (Direct): Banana Gun user → Banana Gun infrastructure → Titan → Validator. Retained margin annotated.
    • Lane 2 (Pipeline): Trader → Phantom wallet → Helius RPC → Jito Block Engine → Validator, with the rebate arrow flowing back from Helius to Phantom.
    • Lane 3 (Validator-side grey market): Searcher → private mempool operator (e.g., DeezNode pre-shutdown) → Validator, with revenue share.

  2. D2 — Validator-quintile decomposition (markdown table). Rows: top-quintile Solana validator, median validator, bottom-quintile validator. Columns: gross SRR, what's known about the access factors (client choice, builder relationships, RPC partnerships), what's known about the operational factors (uptime, latency, hardware), the implied attribution. The empirical anchor for the chapter's central claim — the table demonstrates the order-of-magnitude gap between access and operational drivers.

Glossary terms this chapter introduces ​

To be appended to GLOSSARY.md in Phase 3.

Defined in full (first appearance):

  • Exclusive order flow (on-chain) — a contractual arrangement under which one infrastructure firm provides another party (a searcher, a trading firm, a wallet, a validator) preferential or exclusive access to a slice of trading flow that the firm could otherwise sell to multiple counterparties.
  • Builder-direct relationship — an exclusive flow arrangement specifically between a block builder and a searcher or trading firm, in which the searcher's bundles are routed to the builder under preferential terms.
  • Validator-quintile gap — the empirical observation that the difference in revenue between top-quintile and bottom-quintile validators on most chains is no longer primarily explained by operational excellence but by access to specific infrastructure relationships.
  • Pipeline rebate — an MEV-refund arrangement at the RPC or wallet layer, typically with a contractual split between the infrastructure firm and the wallet customer. Helius's 50/50 split with Phantom / Backpack / Solflare is the canonical example.

Brief mention (one inline sentence; may not warrant full glossary entries):

  • Grey-market private mempool (the DeezNode pattern from Ch 3)
  • HIP-3 builder codes (Hyperliquid's permissioned-but-non-exclusive primitive; full treatment Ch 9)

Backward:

  • Chapter 3 (drafted): the post-Jito-shutdown landscape; DeezNode; the fragmented surfaces. Ch 7 develops the contractual layer Chapter 3 set up.
  • Chapter 4 (drafted): the searcher who pays. Ch 7 develops which searchers pay which infrastructure firms preferentially.
  • Chapter 5 (drafted): the validator role; the client choice as moat. Ch 7's validator-quintile claim builds directly on Ch 5's "moat is the relationship" point.
  • Chapter 6 (drafted): the infrastructure firms. Ch 7 develops which arrangements they actually have with whom.

Forward:

  • Chapter 8 (Solana): chain-specific deep dive that will return to the Solana exclusive-flow patterns developed here.
  • Chapter 9 (Hyperliquid): the chain whose architecture eliminates the surface for these arrangements; HIP-3 builder codes developed in full.
  • Chapter 10 (Ethereum and L2s): the chain where the arrangements are most developed; ePBS as the protocol-level alternative being built.
  • Chapter 11 (Who's at a Disadvantage): the validator-quintile gap as one of the four chronic-loser archetypes.
  • Chapter 12 (Where This Is Going): protocol-level proposals (ePBS, FOCIL, Constellation) that would change the exclusive-flow equilibrium.

Phase 2 is complete. Per the user's compressed-review pattern, Phase 3 begins immediately.