The House Edge

Appearance

Chapter 11 — Research Note (Phase 1) ​

Status: RESEARCH (Phase 1 — written by agent, awaiting Nick's review) Date: 2026-05-14 Working chapter: 11 — Who's at a Disadvantage, and Why They Don't Move Part: IV — The Verdict (Ch 11 opens Part IV; Ch 12 closes the book with forward-looking trends)

How to read this note ​

Ch 11 is the book's "honest reckoning." Its job is to develop four chronic-loser categories across the three architectures Chs 8–10 documented, and then explain why each category continues to participate. The chapter is not a re-litigation of the dollar traces from prior chapters — Ch 3 (Alice's $73), Ch 7 (Banana Gun's 2,271.26 ETH retained), Ch 8 (Alice's compressed ~$8–15), Ch 10 (Alice's ~$15–25 L1 vs ~$2–5 Base) — but a structural treatment of the categorical losers those traces named.

The chapter's load-bearing structural argument is the consent gap: the four loser categories are losers in different senses. A passive LP who deposits expecting fees and gets net-negative LVR is a different kind of loser than a retail memecoin trader who knows the median wallet loses and plays anyway. A small-stake validator who cannot reach the Hyperliquid top-21 threshold of 525,000 HYPE is a different kind of loser than an Ethereum solo staker earning ~4–5% APY versus Lido's 3.5–4% (a winner by raw yield, but a loser by aggregate share). The chapter must engage with consent honestly. Some losers consent to the structure; some don't know what they consent to; some have no available alternative.

Six things to know up front before drafting:

  1. The Pump.fun trader data is the chapter's cleanest retail-loss anchor. Per CoinGecko's May 2026 publication: from April 2024 through late 2025, the share of profitable wallets on Pump.fun rarely exceeded 50%, bottoming at 30.1% in June 2025. In April 2026, 73.3% of wallets ended the month in profit — the highest figure on record — but the recovery was driven by a natural exodus of unprofitable traders: monthly active wallets fell from a peak of 5.2 million in May 2025 to 1.8 million in December 2025 (a ~65% decline). In April 2026, of 3.14 million active wallets, the largest profitable cohort (2.05 million, or 65.1%) earned only $1–$500 for the month. The pattern is the chapter's clearest single illustration of "the retail loser is replaced, not converted": the platform's profitability statistics improved because the chronic losers quit, not because their losses converted to gains. (CoinGecko — Pump.fun Traders Are Making a Comeback; CoinInsider — 73% profitability April 2026; accessed 2026-05-14)

  2. The BIS Bulletin No. 69 is the chapter's retail-aggregate anchor. Bank for International Settlements analysis covering 95 countries and August 2015–December 2022 found that 73–81% of retail crypto-app users likely lost money on their initial Bitcoin investment; the average retail investor lost approximately 47.89% of invested funds. The study's structural finding: as Bitcoin prices rose, smaller users were buying while the largest holders (whales) were selling — a wealth transfer at smaller users' expense. This is the chapter's cleanest cross-chain retail-loss anchor; it is also the most established single piece of evidence in the book that retail crypto users systematically lose money over time, not just on individual trades. (BIS Bulletin No 69 — Crypto shocks and retail losses; BIS — bulletin page; accessed 2026-05-14)

  3. The LVR-net-negative passive LP finding is the chapter's central passive-LP anchor — and it has not been refuted by a 2025–2026 re-measurement. The Topaze Blue + Bancor 2021 study found ~49.5% of measured V3 LPs realised negative returns net of impermanent loss; the Milionis et al. LVR framework formalised the loss as approximately 5–7% of capital per year in volatile pairs from arbitrageurs alone, before fees. The Uniswap Labs 2023 research showed V3 passive LPs in the ETH/USDC 5-bps tier — the highest-volume DeFi pool — underperformed V2 by 68%. No comprehensive 2025–2026 re-measurement of these numbers has been published; the structural finding has not been refuted but has also not been re-validated on fresh data. The chapter must frame this as: the dominant DeFi LP pool is, by the most-cited empirical work, a net loser for the median passive participant, and the structural reason is informed flow against an uninformed pool. (Topaze Blue + Bancor 2021 study via CryptoSlate; Milionis et al. arXiv:2208.06046; Uniswap Labs Research, When Uniswap V3 Returns More Fees for Passive LPs; accessed 2026-05-14)

  4. The slow market maker is named: it's the firm that lost the spot CLOB war, or the firm that wasn't in the original 75 EOF arrangements. Phoenix at $3.7M Q2 2024 → $68,604 Q1 2026 (a 54× decline) is the canonical case (Ch 2, Ch 8). The displaced operator is Ellipsis Labs — they pivoted to Phoenix Perpetuals rather than exit. On Ethereum, the "slow MM" is the firm whose flow does not have a builder-direct contract under Wu et al.'s 75 EOF arrangements (accounting for ~71% of trading-related builder revenue, Ch 7). The losers are not firms exiting visibly; they are firms with structurally lower margin per trade than the prop-AMM operators, the builder-direct firms (SCP/Wintermute at >90% UniswapX, Barter targeting 50%+ CoW), and the Hyperliquid approved counterparties (Wintermute, Flowdesk, FalconX, Nonco — BHYP-named). The displacement is observable in Wintermute's mid-May 2026 90% BTC/ETH liquidity drop on Hyperliquid — even the named winners can rapidly become the named losers when regulatory or operational shocks hit. (DefiLlama Phoenix; Wu et al. arXiv:2405.01329; TechFlow on Wintermute liquidity drop; accessed 2026-05-14)

  5. The small validator is named-and-numbered on all three chains. On Solana: validator count fell from ~2,560 in 2023 to ~770 in March 2026 (a ~70% decline). Approximately 150 validators are projected to lose SFDP foundation stake under the May 2026 rule (<25% ASN, <15% data centre concentration). The Gini coefficient for Solana validator profits is approximately 0.93 (Placeholder VC). On Hyperliquid: slot #21 requires approximately 525,000 HYPE (~$26M at $50/HYPE) to enter the active validator set — even though the self-delegation requirement is only 10,000 HYPE, the practical threshold for an active validator is ~52× higher. On Ethereum: Lido at 24.2% / Coinbase Cloud at 5.1% / Binance at 9.1% of staked ETH means three operators control ~38% of the validator set. Solo home stakers (estimated ~5.4% of staked ETH per the June 2024 ethstaker dataset) earn 4–5% APY versus Lido's 3.5–4%, but their aggregate share is structurally bounded by the 32-ETH minimum. Vitalik Buterin's proposal to drop the minimum to 1 ETH (advocacy, not protocol commitment) is the forward-looking lever. (CCN — Solana 68% validator decline; Phemex — Solana SFDP May 2026; Placeholder VC — Leveling the Stakes on Solana; Hyperliquid docs — staking; cp0x on X — Hyperliquid slot 21 ~525K HYPE; AInvest — Lido vs solo stakers 2026; CoinDesk — Lido share 24.4% August 2025; accessed 2026-05-14)

  6. The "just leave the table" question has a different answer for each loser category, and the chapter's clinical core is making the distinction. Retail can in principle switch to a CEX — but cannot if they are in a jurisdiction where the CEX has been restricted (a meaningful and growing fraction of US users post-2023 Coinbase/Binance enforcement environment), or if their preferred trading venue is memecoin-launchpad style flow that only exists on a permissionless chain. Passive LPs can withdraw — but the alternatives (HODL, stake, stablecoin yield) offer different risk profiles that LPs may have specifically rejected by choosing the pool. Slow market makers can pivot to prop-AMMs — Wintermute did (Tessera V on Solana), Ellipsis Labs did (Phoenix Perpetuals) — but the pivot requires capital, technology, and the operational stack that only the largest firms have. Small validators can exit — and the Solana case (770 from 2,560) shows they do — but exiting is not the same as the network having less concentrated value capture; it just removes the floor of small validators from the distribution while concentrating value further upward. The chapter's structural argument: for three of the four categories, "leaving the table" is available but does not solve the structural problem; for one (small validators), the exit is happening in real time and is the cleanest single piece of empirical evidence that the structural concentration the book has documented is not theoretical.

1. Key claims ​

Each numbered claim is something the chapter is allowed to state. Sources cited inline. The chapter's load-bearing claims are 3 (BIS retail loss), 5 (Pump.fun exodus), 9 (LVR-net-negative passive LP), and 17 (the structural-concentration small-validator-exit pattern).

Retail traders unaware of toxic flow — the first loser category

  1. Alice's 2024 dollar trace updated for the four-chain comparison. The book's named retail anchor across three architectures: ~$73 of 2024 formal-sandwich slippage on Solana (Ch 3); ~$8–15 of 2026 cumulative take on Solana through the full stack (Ch 8); ~$15–25 on Ethereum L1 vs ~$2–5 on Base (Ch 10); ~2–4 bps all-in on Hyperliquid perpetuals (~$2–4 on a $10K notional, Ch 9). The chapter references — does not re-litigate — these traces. The structural finding the chapter develops: Alice's per-trade loss compressed on every chain between 2024 and 2026, but her category-level aggregate loss (Pump.fun's 30–50% profitable-wallet share through 2025) did not. The compression at the trade level is real; the compression at the aggregate level is not. Already cited Chs 3, 8, 9, 10.

  2. Banana Gun's 1.3 million users and 25.3 million executed trades, with $16.09 billion cumulative trading volume and an average trade size of $635, are the chapter's clearest named-retail-category structural anchor. Banana Gun's commercial model is documented in Ch 7: users pay 0.5%–1% per trade on Ethereum, the bot routes exclusively to Titan, Titan retained 2,271.26 ETH of the 4,466.89 ETH user fees (~50.8%), and Titan's builder share rose from <1% to >50% on the strength of that single arrangement. The 1.3M-user count is the under-developed fact in Ch 7's treatment: this is not a marginal product. It is approximately the same order of magnitude as Coinbase's monthly active US retail user base for crypto trading. The structural fact the chapter develops: at least 1.3 million retail users are routing through one bot operator into one exclusive builder contract, paying ~50% of their on-chain costs to a firm whose existence is not disclosed in any product surface the bot's interface displays. (CoinGecko — Top Telegram Trading Bots; Banana Gun research at BlockBase; accessed 2026-05-14)

  3. The BIS Bulletin No. 69 retail-loss aggregate: 73–81% of retail crypto-app users likely lost money on their initial Bitcoin investment; average retail investor loss ~47.89% of invested funds. Coverage: 95 countries, August 2015–December 2022; data on major crypto trading platforms with monthly active users rising from ~100,000 in August 2015 to ~30 million in November 2021. The structural finding: as prices rose, smaller users bought while the largest holders ("whales") sold — making returns at smaller users' expense. The losses are direct and measurable. Crypto's retail loss profile is not a niche phenomenon of memecoin trading or sandwich attacks; it is the structural shape of the asset class as the BIS measured it. (BIS Bulletin No 69; Investment Executive on BIS; Ecofin Agency on BIS 47.89%; accessed 2026-05-14)

  4. The FCA Cryptoasset Consumer Research 2025 (Wave 6) shows UK crypto ownership declined from 12% (2024) to 8% (2025) — from approximately 7 million to ~4.5 million UK users — while average holdings per investor rose to approximately £1,842 (~$2,500). Awareness of cryptoassets remained high (91%), but deeper understanding of stablecoins, DeFi, and crypto lending was variable. 73% of UK users purchased through centralised exchanges (up 4% from 2024). The structural finding the chapter notes: the UK retail market in 2025 is consolidating into a smaller, higher-conviction, CEX-using population — a pattern that does not match the BIS 2015–2022 picture of a growing and rapidly losing population. The compression is jurisdiction-specific and regulator-driven (FCA's MiCA-aligned rules and crypto advertising restrictions). (FCA — Cryptoassets Consumer Research 2025; FCA Wave 6 PDF; CoinDesk on FCA 2025; accessed 2026-05-14)

  5. Pump.fun's profitable-wallet share never crossed 50% from April 2024 through late 2025, bottoming at 30.1% in June 2025; April 2026's 73.3% profitable-wallet share reflects an exodus, not a conversion. Monthly active wallets peaked at 5.2M in May 2025 and fell to 1.8M in December 2025 (a ~65% decline). In April 2026: 3.14M active wallets, 2.30M profitable, with the largest profitable cohort (2.05M, or 65.1% of profitable wallets) earning only $1–$500 for the month. 793,000 wallets (~25% of active) lost between $1 and $500 in the median month; 46,000 wallets (~1.5%) lost more than $500. The structural finding: Pump.fun's improving aggregate statistics reflect the structural shape Ch 11 documents — chronic losers exit, the platform looks healthier, but the loss rate per new entrant has not been measured to have improved at the entry stage. (CoinGecko — Pump.fun Traders Are Making a Comeback; CoinInsider — 73% profitability April 2026; Phemex — CoinGecko 73.3%; accessed 2026-05-14)

  6. The prospect-theory framing of retail crypto: lottery-like positively-skewed return distributions are the documented attraction for the retail trader. The academic literature (Thoma; Aharon; Salience Theory of Cryptocurrency Returns) is consistent: high prospect-theory-value cryptocurrencies are highly positively skewed, and investors overweight the probability of extreme positive outcomes (the "lottery ticket" effect). FOMO, herd behaviour, and the "low entry cost → high reward illusion" are the named psychological drivers. The chapter uses this literature to develop the consent argument: a retail trader who consciously buys a memecoin as a lottery ticket is participating in a structure she understands the asymmetry of; a retail trader who routes a USDC→SOL swap through Phantom + Jupiter without knowing about prop-AMMs and toxic flow is participating in a structure she does not. The two losers are different. (Thoma — Prospect Theory and Crypto Returns; AInvest — Meme Coin Volatility psychological drivers; Salience Theory — University of Liverpool; accessed 2026-05-14)

Passive LPs in informed-flow pools — the second loser category

  1. The Loss-Versus-Rebalancing (LVR) framework is the chapter's central passive-LP anchor. Milionis, Moallemi, Roughgarden, and Zhang's 2022 paper (arXiv:2208.06046) formalised the LP problem as the difference between the LP's P&L and a constant-mix rebalancing portfolio's P&L. The key result: an LP in an AMM with 5% daily price volatility loses approximately σ²/8 per unit time — which annualises to approximately 11% per year to arbitrageurs alone, before fees. ETH-USDC and BTC-USDC pairs lost approximately 5–7% of capital per year to LVR in 2023. The structural finding: passive LPs in volatile pairs cannot break even on LVR unless fee yields exceed the LVR rate, which the empirical data shows they generally do not. (Milionis et al. — arXiv:2208.06046; CoW — What is LVR; a16z crypto on LVR; accessed 2026-05-14)

  2. The Topaze Blue + Bancor 2021 study: 49.5% of measured V3 LPs realised negative returns net of impermanent loss across 17 V3 pools (~43% of V3 TVL). $199.3M in collected fees against $260.1M in IL — passive LPs were collectively ~$60.8M worse off than HODL. No comprehensive 2025–2026 re-measurement of this finding has been published. A 2025 MEXC Research commentary (circulating in secondary citations) puts the share of LPs in volatile pairs losing money net of fees at ~54.7% — directionally consistent. The structural finding: half of V3 passive LPs are worse off than holding the underlying assets, before LVR even enters the picture. Ch 11 cites this as the chapter's structural-loser anchor for passive LPs; the empirical work has not been refuted by any comparable published 2025–2026 study. Already cited Ch 2. (CryptoSlate on Topaze Blue study; accessed 2026-05-14)

  3. The Uniswap Labs 2023 research: V3 passive LPs in the ETH/USDC 5-bps tier — the highest-volume DeFi pool — underperformed V2 by 68%. Across all tiers and pools studied, V3 passive LPs earned higher fees than V2 LPs by an average of about 54%; but in the dominant ETH/USDC 5-bps tier, the most-traded pool is the worst case for the passive LP. The mechanism: the most-traded pool is the pool where counterparties most often know more than the pool does, and the passive LP absorbs adverse selection at the LP's expense. The chapter's structural argument: the dominant DeFi LP pool is, by Uniswap's own research, where passive LPs lose worst. Already cited Ch 2. (Uniswap Labs Research — When Uniswap V3 Returns More Fees for Passive LPs; accessed 2026-05-14)

  4. Hyperliquid's HLP is the structurally-different passive-LP-class equivalent — and JELLY (26 March 2025) demonstrated the tail risk for them too. HLP TVL fell from approximately $500M (pre-12 March 2025) to ~$177M by 1 April 2025 (a 65% drawdown) on the back of a single targeted whale attack on the JELLY perpetual. The Hyper Foundation delisted JELLY and closed positions at $0.0095 (a $700K gain to HLP); HLP recovered to ~$383M TVL by May 2026 (Ch 9). Historical HLP returns: ~1.75% per month / ~20% annualised through 2025; a $700M February 2026 liquidation generated approximately $15M in HLP profit. The HLP is structurally distinct from a Uniswap V3 passive LP — the depositor knows she is funding the chain's house and absorbs both upside (97% of validator-perp fee residual through the Assistance Fund) and tail risk — but the category-level loser dynamic is comparable: any HLP depositor participating during a single JELLY-style event eats a fraction of the drawdown. The chapter must develop HLP as the cleanest case of a passive-LP-equivalent whose risk profile is more disclosed than Uniswap V3's. (KuCoin — HLP Vault explainer; ARX — Hyperliquid Vaults Explained; Ch 9 RESEARCH.md claim 4; accessed 2026-05-14)

  5. The toxic flow / informed flow asymmetry is the structural mechanism underlying the passive-LP loss. Arbitrage order flow is "toxic" (forces the LP to buy or sell against the market); other flow is "uninformed" (mostly retail). Two clusters of negative-PnL wallets with high notional swap sizes have been identified by CrocSwap's series on toxic flow detection — and aggregator-routed flow (Jupiter, CoW Swap) appears to remain a credible signal of non-toxicity not easily gameable by arbitrageurs. The chapter's structural finding: passive LPs in the most-traded pool absorb the most-toxic flow; the solver markets and aggregators have routed retail flow away from passive pools and toward prop-AMMs and PMM solver inventory. The passive LP is the structural residual — left holding the toxic component because the non-toxic retail component was siphoned off by the aggregators. (CrocSwap — Discrimination of Toxic Flow in Uniswap V3 Part 4; Sandmark — Toxic Flow: The Hidden Cost; accessed 2026-05-14)

Slow market makers — the third loser category

  1. Phoenix's $3.7M Q2 2024 → $68,604 Q1 2026 — a ~54× revenue collapse — is the canonical "MM that lost the spot CLOB war" case. Ellipsis Labs renamed Phoenix to Phoenix Legacy and pivoted to perpetuals (Phoenix Perpetuals, announced Breakpoint 2025). The displacement was not a single firm exiting; it was the entire competitive form (on-chain spot CLOBs on Solana) failing to compete with prop-AMMs and aggregator-routed flow. Manifest at ~$3.9B 30-day spot volume captures <2% of the $284.5B Q1 2026 Solana DEX volume; OpenBook is in the same compressed band. The slow MM on Solana is not a named firm — it is the category of market makers whose business depended on tight-spread quoting against a public CLOB and who could not compete with the prop-AMM internalisation that ate their flow. Already cited Chs 2, 8. (DefiLlama Phoenix; Ellipsis Labs — Introducing Phoenix Perpetuals; accessed 2026-05-14)

  2. The Ethereum slow MM is the firm that did not capture a builder-direct contract under Wu et al.'s 75 EOF arrangements (accounting for ~71% of trading-related Ethereum builder revenue). Pahari & Canidio's December 2024 measurement: 77.2%–84% of total fees paid in winning Ethereum blocks come from exclusive transactions not available through public order flow. The OG L1 market makers — the firms that quoted into the public mempool and pre-MEV-Boost-private-order-flow architecture — have either pivoted (Beaverbuild migrated into BuilderNet; Flashbots stopped operating a centralised builder December 2024) or compressed (smaller firms that did not have the capital to build a builder operation themselves). The 2026 winners are SCP and Wintermute (>90% of UniswapX volume between two firms); Barter (targeting 50%+ of CoW solver share); the named approved counterparties for the regulated venues (Bitwise BHYP: Wintermute, Flowdesk, FalconX, Nonco). Already cited Chs 2, 5, 6, 7. (Wu et al. — arXiv 2405.01329; Pahari & Canidio — arXiv 2509.16052; accessed 2026-05-14)

  3. The Solana market makers without Frankfurt/Equinix FR5 colocation, BAM Node access, or Harmonic strategy enablement earn at-or-below network median priority fees — the +101% Harmonic Performance gap measured against the median is the chapter's "speed-and-relationship moat" anchor. Syndica March 2026 measurements: Frankendancer Harmonic Performance +101%, Frankendancer Harmonic Balanced +39%, Agave Harmonic +36% per-block priority fees vs network median. Chorus One's August 2025 finding: combined operational gains (timing games + scheduler optimisation) amount to ~+3% in total rewards (~27 bps annualised). The ratio is approximately an order of magnitude — and the slow MM on Solana is the operator at network median or below, by definition. Already cited Chs 5, 7, 8. (Syndica — March 2026 deep dive; Chorus One — Timing Games on Solana; accessed 2026-05-14)

  4. The mid-May 2026 Wintermute liquidity drop on Hyperliquid demonstrates that the winners can become losers fast. Per TechFlow's coverage (18 May 2026): Wintermute and Auros Global simultaneously withdrew ~$100M of aggregate liquidity from Hyperliquid; Wintermute's combined BTC + ETH liquidity dropped from ~$40M to ~$4M (a ~90% decline). The withdrawal occurred three days after CME and ICE jointly pressured US regulators to review Hyperliquid — the structural reading is that even the named approved trading counterparties (the Bitwise BHYP set) reduce exposure when regulatory uncertainty rises. The slow-MM-as-loser frame must include this: the winners are not stable; the regulatory and operational risk profile of being a top-tier on-chain MM in 2026 has grown materially even for the firms that have the most exclusive access. (TechFlow — BTC Liquidity Plummets 90%; accessed 2026-05-14)

Validators without infrastructure relationships — the fourth loser category

  1. The access-vs-operational decomposition on Solana: +33%–+101% access signal vs ~+3% operational signal — the validator-quintile gap is access-driven, by approximately an order of magnitude. Syndica's March 2026 deep dive measured per-block priority fees vs network median: Frankendancer Harmonic Performance +101%, Frankendancer Harmonic Balanced +39%, Agave Harmonic +36%. Chorus One's August 2025 measurement: combined timing-games + scheduler optimisation produced +3.0% in total rewards (~27 bps annualised). The ratio is ~10× — meaning a validator that exhausts every operational optimisation publicly known still earns materially less than a validator with the right relationships and client stack. Already cited Chs 5, 7. The chapter's loser category here is the bottom-quintile Solana validator — running stock Agave, no SFDP delegation, no BAM Node relationship, no Frankfurt colocation, no Harmonic strategy. The Placeholder VC measurement: Gini coefficient for Solana validator profits ~0.93 (extreme inequality; more unequal than the most-unequal US states or the most-unequal countries globally). (Syndica; Chorus One; Placeholder VC — Leveling the Stakes on Solana; accessed 2026-05-14)

  2. The Solana validator exit pattern is real and named: ~770 validators in March 2026, down from ~2,560 in 2023 — a ~70% decline. Approximately 150 validators are projected to lose SFDP foundation stake under the May 2026 rule (<25% ASN, <15% data centre concentration). The chapter's structural reading: the small-validator loser category is not theoretical; it is exiting in real time. The Foundation's "onboard 1, offboard 3" SFDP policy is the explicit mechanism. The remaining 770 are heavily concentrated in the named Equinix FR5 / Amsterdam / London anchor data centres (Ch 8: Frankfurt 19%, Amsterdam 16%, London 12% of stake). The validator exit is the cleanest single piece of empirical evidence in the book that the structural concentration the chapters have documented is producing observable participant-loss. (CCN — Solana 68% validator decline; Phemex — SFDP May 2026 rules; Blockworks — Solana Foundation begins pruning validators; Helius — Solana Foundation Delegation Program & Challenges for Validators; accessed 2026-05-14)

  3. The Hyperliquid small-stake validator: slot #21 requires ~525,000 HYPE (~$26M at $50/HYPE), even though self-delegation is only 10,000 HYPE. The set is mid-expansion from 24 to 27 (Foundation announcement, 18 May 2026). The structural fact: any operator can register a validator with 10,000 HYPE self-delegation locked for one year; only the top 24 (expanding to 27) by total delegated stake form the active set. A validator entering "undelegate-only mode" if self-delegation drops below threshold. The 525,000 HYPE / 10,000 HYPE ratio is ~52× — a small-stake operator with the technical capacity to run a validator faces a stake-acquisition barrier ~52× her self-delegation. This is a different shape from Solana's (the floor is operational economics) and Ethereum's (the floor is the 32 ETH minimum, which Vitalik has proposed lowering to 1 ETH). On Hyperliquid the floor is delegated stake competition — a permissionless-but-zero-sum competition for the 21-to-27 active slots. Already cited Ch 5, Ch 9. (Hyperliquid Docs — Staking; cp0x on X — slot 21 ~525K HYPE; Coin Edition — Hyperliquid validators 24→27; accessed 2026-05-14)

  4. The Ethereum solo home staker: ~5.4% of staked ETH (per ethstaker June 2024 dataset); earns 4–5% APY vs Lido's 3.5–4%; structurally bounded by the 32 ETH minimum. Lido at 24.2% of staked ETH (8.7M ETH); Coinbase Cloud at 5.1% (1.84M ETH); Binance at 9.1% (3.29M ETH) — three operators control ~38% of the validator set. The chapter's structural argument for Ethereum: the solo staker is a winner by raw yield (Lido takes a ~10% fee; solo stakers keep all of it), but a loser by aggregate share and by the asymmetric access to MEV-Boost optimal-relay routing, builder relationships, and the institutional staking-as-a-service infrastructure (Coinbase Cloud, Kiln, Figment, Lido's curated operator set). Vitalik Buterin's advocacy for lowering the staking minimum to 1 ETH is not a protocol commitment; the structural shape is set. Already cited Ch 5. (CoinDesk — Lido market share August 2025; Datawallet — Ethereum Staking Statistics 2026; AInvest — solo stakers 4-5% APY; Vitalik — solo staking advocacy; ethstaker — solo-stakers dataset; accessed 2026-05-14)

Why each group continues to participate anyway — the "why they don't leave the table" mechanism

  1. Retail's switching cost: jurisdictional CEX access plus UX gravity. The DEX-to-CEX spot volume ratio more than tripled 2021–2025 (6.0% → 21.2% as of November 2025); spot DEX share reached ~14% of global spot in January 2026 after peaking at 24.5% mid-2025. The shift is documented and is in the opposite direction of "retail moves to safer venues." Retail is shifting to DEXs, not away. The chapter's structural reading: the alternatives retail has are (a) a CEX in a jurisdiction where it is available (the BIS finding documents 95-country coverage, but US users post-2023 face progressively restricted CEX options), (b) self-custody on a DEX (Phantom + Jupiter, Coinbase Wallet + Aerodrome), or (c) exit the asset class entirely (the FCA-observed 4M UK user decline 2024→2025). Retail is moving toward (b) and partly toward (c); option (a) is structurally bounded. (CCN — Crypto Users Shifting from CEX to DEX; CoinGecko — DEX to CEX Volume Ratios; FCA 2025 research; accessed 2026-05-14)

  2. The PFOF regulatory asymmetry as the chapter's TradFi parallel for "why each group continues to participate" — the EU bans, the US permits, the on-chain version has neither. EU PFOF ban effective 30 June 2026; only Germany exercised the transitional exemption. SEC withdrew the Order Competition Rule on 12 June 2025 and Regulation Best Execution. The chapter's structural reading: a retail trader in the EU has explicit regulatory protection against the exchange-of-execution-quality-for-rebates pattern that exists in US equity markets; the same retail trader in the US does not; the same retail trader on Solana through Phantom + Helius's 50/50 wallet rebate has neither. The chapter references — does not re-litigate — Ch 7's regulatory framing. Already cited Ch 7. (ESMA — PFOF exemption list; SEC withdrawal of rule proposals 12 June 2025; accessed 2026-05-14)

  3. MiCA's EU-wide enforcement (1 July 2026) adds the second layer of EU-side regulatory protection for retail — the 14-day cooling-off period for direct cryptoasset purchases from the offeror. No equivalent exists in the US under the current SEC. The chapter develops MiCA as the closest TradFi-style retail-execution-protection regime available to an EU crypto trader in 2026 — and notes that MiCA does not address trade execution per se; it addresses asset offering and consumer disclosure. The structural gap remains: a retail trader's per-trade execution quality is unregulated on every chain in every jurisdiction the book has documented. (Sumsub — MiCA Guide 2026; ESMA — MiCA activities; accessed 2026-05-14)

  4. Passive LPs continue depositing because the advertised APYs survive the LVR-net-negative finding — the Uniswap V3 5-bps tier advertises fee yields that, in isolation, exceed the LVR rate; but the LVR loss is not a line item the LP-deposit interface displays. The chapter's structural finding: the "you're an LP, you earn fees" framing the V2 design popularised has survived the V3 design's empirical reversal in the most-traded pool. No DEX in 2026 displays an LVR-adjusted P&L estimator at the point of deposit. Solutions exist at the protocol level (CoW AMM batch auctions; Panoptic; oracle-based AMM designs) but none has replaced the dominant V3-style design. The structural fact: passive LPs continue to deposit, in 2026, into pools whose LVR-net loss has been published in peer-reviewed work for four years.

  5. Slow MMs continue to participate because pivoting to a prop-AMM requires capital, technology, and the operational stack only the largest firms have. Wintermute (Tessera V on Solana) and Ellipsis Labs (Phoenix Perpetuals) are the named pivoters. The smaller firms — the ones running standalone quoting bots on Phoenix, OpenBook, or any of the on-chain CLOBs that compressed — do not have the option to internalise prop-AMM economics because the operational stack (closed pricing engine, on-chain settlement infrastructure, aggregator-routing relationships) requires multi-million-dollar capital outlay. The chapter's structural argument: the "just pivot" answer to slow MMs is available to the largest 5–10 firms. For the rest, the exit is to non-crypto trading or to a smaller crypto venue with worse economics.

  6. Small validators continue to operate because the alternative is selling the operational sunk cost. The Solana 770 → ~620 (estimated post-SFDP-removal) trajectory documents the rate at which they don't continue. On Hyperliquid, the small-stake operator either accumulates 525K HYPE in delegation (which requires marketing, reputational capital, and time) or runs in "undelegate-only mode" — not actively staked but technically a validator. On Ethereum, the solo staker's 4–5% APY vs Lido's 3.5–4% gives a positive incremental yield, but the absolute differential on 32 ETH at $3,500/ETH (= $112,000 capital) is ~$560–$1,680/year on a single validator — a meaningful but small reward for the operational complexity. The chapter develops the per-chain answer: on Solana the exit is fast and visible; on Hyperliquid the entry is structurally bounded; on Ethereum the participation is structurally subsidised by the protocol's solo-staker advocacy. The shapes are different. Already cited Ch 5.

The consent gap — the chapter's structural moral argument

  1. The four loser categories differ in their informed-consent relationship to the structure. The chapter's clinical core. A passive LP who expects to be an LP at a structural disadvantage (sophisticated trader using AMMs as inventory) is not the same as a passive LP who believes they're earning fees from uniformly-priced flow. A retail memecoin trader who consciously buys a lottery ticket (prospect-theory consent) is not the same as a retail USDC→SOL swapper who does not know Helius's 50/50 wallet rebate is being captured at the RPC layer. A market maker who chooses to pivot to prop-AMMs has different agency than a market maker whose business failed because the spot CLOB design lost the market. A solo Ethereum staker who chooses the 4–5% APY over Lido's 3.5–4% has different consent than a Solana validator delegated to the SFDP whose stake is removed by the May 2026 rule change. The chapter must make all four distinctions explicitly, without moralising, and without framing the structural fact (concentration produces categorical losers) as a function of any single actor's bad faith. The losers' loss is structural; the losers' consent is heterogeneous. The chapter's job is to develop the consent profile per category, then leave the policy question to Ch 12.

2. Numbers to verify before drafting ​

Numbers from prior chapters are referenced — not re-litigated — and tagged "✓ prior" below. New numbers are flagged for verification.

Retail loss anchors

  • Alice's dollar traces: ~$73 (2024, Ch 3) / ~$8–15 (2026 Solana, Ch 8) / ~$15–25 L1 vs ~$2–5 Base (Ch 10) / ~$2–4 Hyperliquid perp (Ch 9) — ✓ prior.
  • BIS 73–81% retail loss / 47.89% average loss — published, August 2015–December 2022 window; 2025–2026 update has not been published. Flag as stale window (verify with Nick whether to use). BIS Bulletin 69
  • FCA UK crypto ownership 12% → 8% (2024 → 2025); ~7M → ~4.5M users; mean holding ~£1,842 — 2025 Wave 6 PDF. ✓ recent.
  • Pump.fun: 30.1% profitable wallets June 2025; 73.3% April 2026; 5.2M → 1.8M monthly active wallets (May 2025 → December 2025); 3.14M April 2026 / 2.30M profitable / 65.1% of profitable earned $1–$500. ✓ recent (CoinGecko May 2026).
  • Banana Gun: 1.3M users, 25.3M trades, $16.09B volume, avg trade $635. ✓ recent (CoinGecko Telegram bot review).

Passive LP anchors

  • Topaze + Bancor 2021: 49.5% V3 LPs net-negative; $199.3M fees / $260.1M IL / $60.8M aggregate net loss across 17 pools (43% of V3 TVL). ✓ prior (Ch 2). Flag: 2021 dataset; no 2025-2026 re-measurement.
  • LVR: σ²/8 per unit time; ~11% annualised at 5% daily volatility; ~5–7% capital/year ETH-USDC / BTC-USDC 2023. ✓ prior (Ch 2).
  • Uniswap Labs 2023: V3 +54% fees vs V2 on average; V3 −68% in ETH/USDC 5-bps tier. ✓ prior (Ch 2).
  • HLP: $500M (pre-12 Mar 2025) → $177M (1 Apr 2025) → $383M (May 2026); JELLY 12M unrealised loss / 700K close-out gain; ~$15M Feb 2026 liquidation gain. ✓ prior (Ch 9).

Slow MM anchors

  • Phoenix Q2 2024 $3.7M → Q1 2026 $68,604 (54× decline). ✓ prior (Chs 2, 8).
  • Wu et al. 75 EOF arrangements / ~71% of trading-related Ethereum builder revenue. ✓ prior (Ch 7).
  • Pahari & Canidio 77.2%–84% of fees in winning blocks from exclusive transactions. ✓ prior (Ch 7).
  • Wintermute mid-May 2026 BTC + ETH liquidity 90% drop (~$40M → ~$4M). ✓ recent (TechFlow).

Small validator anchors

  • Solana validator count 2,560 (2023) → 770 (March 2026); ~150 projected SFDP removals May 2026. ✓ recent (CCN; Phemex). Flag: 770 figure is a single source; cross-validate with Solana Foundation data if available.
  • Solana validator profit Gini ~0.93 (Placeholder VC, September 2025). ✓ prior (Ch 7).
  • Solana access vs operational signal: +33% to +101% vs +3%. ✓ prior (Chs 5, 7, 8).
  • Hyperliquid slot #21 ~525K HYPE; 10K HYPE self-delegation; 24 → 27 expansion. ✓ prior (Ch 9).
  • Lido 24.2% / Coinbase Cloud 5.1% / Binance 9.1% of staked ETH. ✓ prior (Ch 5).
  • Ethereum solo stakers ~5.4% of staked ETH (June 2024 ethstaker dataset). Flag: dataset is ~2 years old; no comprehensive 2026 update.
  • Solo staker APY 4–5% vs Lido 3.5–4% (post-Lido-fee). ✓ recent (AInvest, Spotedcrypto).
  • Solana SFDP rules: <25% ASN, <15% data centre. ✓ prior (Ch 8).

Regulatory anchors

  • EU PFOF ban 30 June 2026; only Germany exempted. ✓ prior (Ch 7).
  • SEC withdrawal of Order Competition Rule 12 June 2025. ✓ prior (Ch 7).
  • MiCA full enforcement 1 July 2026; 14-day retail cooling-off period. ✓ recent.

DEX/CEX retail flow

  • DEX:CEX spot ratio 6.0% → 21.2% (Jan 2021 → Nov 2025); spot DEX share ~14% of global spot Jan 2026 (peaked 24.5% mid-2025). ✓ recent (CoinGecko).

3. Contested or evolving claims ​

  1. Retail-loss-rate update post-2022. The BIS 2015–2022 73–81% retail loss finding is the most-cited single piece of empirical work in the book on retail crypto loss. The window ends December 2022 — before the 2023–2024 memecoin cycle, before MiCA enforcement, before the FCA's crypto promotion crackdown, before Pump.fun. The 2026 picture (Pump.fun 30–50% profitable for 18 months, then a ~65% wallet exodus, then 73% profitable on a smaller base) may be either consistent with the BIS finding (chronic losers exit; survivors look better) or evidence of an improved retail-loss profile per-trade with a higher-conviction smaller surviving population. Open question for Nick: is there a 2024–2026 update to the BIS finding I should be citing?

  2. LVR re-measurement post-2023. The Milionis et al. and Topaze Blue empirical work is the chapter's central passive-LP loss anchor, and the most recent comprehensive measurement is from 2021–2023. The 2024–2026 landscape — V4 hooks, solver-routed retail flow that may not reach passive pools, prop-AMM displacement — may have changed the passive-LP P&L profile materially. No comprehensive 2025–2026 re-measurement has been published that I located. Flag: this is the chapter's most-cited single empirical anchor and it is 2–4 years old. The chapter should cite as such and acknowledge the staleness.

  3. The Banana Gun 1.3M user count vs the empirical retail-user-loss profile is asymmetric: the chapter has a published user count but no published per-user P&L distribution. Pump.fun has the P&L distribution (CoinGecko). Banana Gun does not. The chapter can describe the structural fact (Banana Gun users pay Titan ~50% of their fees) and the user count, but cannot make a parallel claim to Pump.fun's 30–73% profitable-wallet shape. Flag for verification with Nick whether the Banana Gun framing should claim or stop short of claiming a comparable loss profile.

  4. HLP "loser" framing. The HLP is the chain's house and is, by design, structurally distinct from a Uniswap V3 passive LP. The chapter must not conflate the two. A passive V3 LP loses by being on the wrong side of informed flow without consent or disclosure; an HLP depositor signs up for the "you fund the chain's worst flow" position and earns ~20% annualised on the back of that role. The JELLY incident demonstrates the tail risk, but the HLP depositor was on the paid side of the JELLY trade after the Foundation's delisting. The chapter's framing must be careful here. Recommendation: develop HLP as a structurally-different-loser-class rather than as another instance of the passive-LP loser.

  5. Ethereum solo staker as winner vs loser. A solo home staker earns 4–5% APY vs Lido's 3.5–4% — by raw yield she is a winner. By aggregate share (~5.4% of staked ETH; declining as Pectra consolidation favours institutional operators), she is a loser. By access to MEV-Boost optimal-relay routing, builder-direct relationships, and the institutional staking-as-a-service infrastructure, she is a loser. The chapter must frame this dual-status carefully. Recommendation: frame as "the solo staker is a winner per validator slot and a loser as a structural participant — the structural concentration is happening above her, not at her."

  6. The "leave the table" answer for retail in the US. The DEX:CEX ratio is moving in favour of DEXs, but the chapter's outline asks whether retail education would change the loss profile. The empirical answer is: not measurably. FCA crypto disclosure rules in the UK reduced ownership from 12% to 8% — but the surviving 8% has higher average holdings, not lower per-trade losses. The MiCA cooling-off period is 14 days for direct cryptoasset purchases from the offeror — but trade execution per se is not covered. The structural finding is that consumer protection regimes can reduce participation but have not been demonstrated to change per-trade loss profiles for active participants. Flag for the chapter to acknowledge this honestly rather than imply that more disclosure would help.

  7. Wintermute on Hyperliquid: winner or loser? The 90% liquidity drop on 18 May 2026 looks like a loss; but it's structurally a de-risking response to regulatory pressure (CME/ICE letter three days prior). The same firm is named as a Bitwise BHYP approved counterparty — and is a known winner in the SCP-Wintermute UniswapX duopoly. The chapter must frame Wintermute as winner-in-equilibrium-but-loser-in-shock — and develop the structural reading that even the named winners have non-trivial drawdown risk when regulatory or operational shocks hit. Recommendation: develop this as Ch 11's clearest case that "winner" status is conditional on a stable regulatory environment, and that the regulatory environment in 2026 is not stable.

4. Characters introduced ​

First-introduced-in-Ch-11

  • The Pump.fun cohort. Not a named individual; a category. The 5.2M-peak/1.8M-trough/3.14M-recovery monthly active wallet population on the Solana memecoin launchpad. The cleanest empirical anchor for the chapter's retail-loser category. No character development needed — the dataset is the character.

  • The bottom-quintile Solana validator. Not a named individual; a category. The operator running stock Agave, no SFDP delegation, no Frankfurt colocation, no Harmonic strategy. The chapter develops this as the cleanest single case of the access-gap loser. Cross-reference: the firms named in prior chapters as the winners (Helius, Triton One, SOL Strategies, Figment for BAM; Frankendancer Harmonic Performance operators) implicitly identify this loser as the validator not in those names.

  • The slot-#22 Hyperliquid validator. The hypothetical operator with technical capacity but insufficient delegated stake to reach the 525K HYPE threshold. The chapter develops as the cleanest single case of the permissionless-but-stake-gated loser — different from a Solana SFDP-removed validator and different from an Ethereum solo staker.

  • The Ethereum solo home staker. ~5.4% of staked ETH per ethstaker; the operator running her own 32-ETH node in her living room, earning 4–5% APY vs Lido's 3.5–4%. Vitalik Buterin's advocacy character but never the protocol's structural commitment.

Returning from prior chapters

  • Alice (retail). The book's protagonist. The chapter does not give her a new dollar trace — it references Chs 3, 8, 9, 10. The character returns as the retail-loser anchor and as the chapter's lens on the consent gap. The chapter's question is not "what did Alice pay" but "what did Alice know about what she paid, and would knowing have changed her behaviour?"

  • Bob (LP). Chapter 2's $50,000 Uniswap V3 ETH/USDC LP. The chapter returns Bob as the passive-LP-loser anchor. His monthly P&L (Ch 2): ~$350 fees collected, ~$475 LVR loss, net -$125. The chapter develops the structural question: why is Bob still an LP in May 2026?

  • Carlos (institutional Ethereum trader). Chapter 3's institutional trader using Flashbots Protect. The chapter notes Carlos is not the loser category — he has the awareness and operational sophistication to route around the public surface. The chapter uses Carlos as the counterfactual: what does Alice's loss look like if she had Carlos's operational capability?

  • Coinbase Cloud (validator). Ch 5's Ethereum validator anchor. The chapter returns Coinbase Cloud as the winning validator (5.1% of staked ETH; institutional staking-as-a-service; named in Lido's curated operator set) against which the solo staker is structurally bounded.

  • Wintermute / Flowdesk / SCP / Barter (market makers). Chs 2, 5, 7 named the winning MM firms. The chapter returns them as the comparison against which the slow MM is defined.

  • Titan, BuilderNet, Quasar, Helius, Jito Labs (infrastructure firms). Chs 5, 6, 7. The chapter returns them only by reference; the chapter's job is not to re-litigate the firms' winning positions.

  • Phoenix (CLOB), HumidiFi/Tessera/SolFi/BisonFi (prop-AMMs). Chs 2, 8. The chapter references the spot CLOB decline as the slow-MM-anchor case.

5. Worked example candidates ​

The chapter's worked-example structure is different from Chs 1–10's. The prior chapters threaded one worked example (Alice's $10K swap; Bob's $50K LP) through the chapter. Ch 11's structural job — developing four loser categories across three architectures — argues for four short worked examples rather than one long one. The Goldman MD test: she wants to see one case per category, not one case spread across all four.

Candidate A — The Pump.fun chronic loser (RECOMMENDED). The chapter's strongest single anchor. A retail trader on Pump.fun in June 2025, when the platform's profitable-wallet share was 30.1%. She executes ~20 trades per month; 60–70% of them lose; her aggregate monthly P&L is ~−$200 on a $3,000 deployed balance. She exits in August 2025, joining the 65% population decline from 5.2M to 1.8M monthly actives. The platform's profitability statistics improve to 73% in April 2026 — not because her cohort converted to winners, but because her cohort left. The example does two things: (a) names the retail loser at the per-trader-per-month level with published data; (b) demonstrates the structural pattern (chronic losers exit; the platform looks healthier; the loss rate per new entrant has not been measured). The chapter's strongest single case. Recommend as the lead worked example.

Candidate B — Bob's LP P&L revisited. Returns Bob from Ch 2. $50,000 in the V3 ETH/USDC 5-bps tier; $350 monthly fees; $475 monthly LVR; net −$125/month; ~3% annualised drag. The chapter develops the question: why is Bob still an LP in May 2026? The structural answer: because the 5-bps tier's published APY (annualised fee yield) at 7%+ exceeds the user-visible alternatives (USDC stable yield 4–5%; ETH staking 3.5–4%); the LVR loss is not displayed at deposit time; and Bob's individual P&L is dominated by ETH's price movement, not the LVR drag. The chapter develops this as the cleanest unaware participation case. Recommend as the secondary worked example.

Candidate C — The Solana SFDP-removed validator. A small Solana operator running stock Agave in a single data centre with ~700 SOL of foundation delegation through SFDP. On 1 May 2026, the operator falls outside the new <25% ASN / <15% data centre concentration rules and loses the foundation delegation. The operator's external stake (~50 SOL) is insufficient to cover operational costs; the operator exits within 60 days. The example develops the small-validator-as-loser at the category level — one of the ~150 projected removals. The chapter develops this against the comparison to the Frankendancer Harmonic Performance operator running in Equinix FR5 with +101% per-block priority fees. The contrast is the chapter's clearest single case of the access-driven Gini ~0.93 concentration. Recommend as the third worked example.

Candidate D — The Wintermute mid-May 2026 liquidity drop. A named firm, a documented 90% liquidity reduction in BTC + ETH on Hyperliquid, occurring three days after CME/ICE pressured US regulators. Develops the slow-MM-as-loser frame asymmetrically: the winner (Wintermute is in the BHYP-approved counterparty set, the SCP-Wintermute UniswapX duopoly, and Tessera V on Solana) becomes the temporary loser when regulatory shock hits. The chapter develops the structural reading: even the named winners have material drawdown risk. Optional fourth worked example; the chapter's case for "winner status is conditional." Not strictly necessary if the chapter develops the slow MM category at the category level rather than the firm level.

Recommendation. Three short worked examples — A (Pump.fun chronic loser as the lead), B (Bob revisited as the passive-LP loser), C (the SFDP-removed validator). One per category, with the slow-MM category developed at the category level rather than through a specific worked example. The chapter's strongest structural shape: open with the Pump.fun cohort data, develop the four loser categories with the three worked examples threaded across them, and close with the "leave the table" structural reading per category.

6. Open questions for Nick ​

  1. The 2024–2026 retail-loss update. Is there a post-BIS-Bulletin-69 retail-crypto-loss empirical study I should be citing? The BIS data ends December 2022; the FCA UK Wave 6 covers 2025 ownership but not per-trade loss profiles; the CoinGecko Pump.fun data is platform-specific. Is there a cross-platform, cross-chain 2024–2026 retail-crypto-loss measurement I'm missing?

  2. The Banana Gun loss-distribution data. I have Banana Gun's user count (1.3M), trade count (25.3M), volume ($16.09B), and average trade size ($635). I do not have a published per-user P&L distribution comparable to the CoinGecko Pump.fun data. Is the chapter expected to develop the Banana Gun retail-loser case at the user-loss-distribution level, or only at the cumulative-fee-to-Titan-2,271.26-ETH level?

  3. The HLP-as-loser framing. The HLP is structurally distinct from a Uniswap V3 passive LP (the depositor knows what she funds; the role is the chain's house; the JELLY incident was a stress test, not a normal-state P&L event). Does the chapter develop HLP as a passive-LP loser, as a separate category (a "structurally-disclosed informed-flow absorber"), or as a winner-with-tail-risk? My current recommendation is the second — HLP as a structurally-different-loser-class — but it's a judgement call.

  4. The Ethereum solo staker dual-status. A solo staker is a winner by raw yield (4–5% vs Lido's 3.5–4%) and a loser by aggregate share (~5.4% and declining). The chapter must frame this carefully — recommendation: "the solo staker is a winner per validator slot, a structural participant in a system whose concentration is happening above her" — but the framing is the chapter's most delicate single sentence and I would value Nick's review of the framing before drafting.

  5. The chapter's stance on "would regulation help." The outline asks whether retail education would change retail's loss profile (cite empirical studies if they exist; flag if not). The empirical answer is: FCA's crypto restrictions reduced UK ownership from 12% to 8% but the surviving population has higher average holdings, not lower per-trade losses. Should the chapter take a clinical stance on this, or stop at "the empirical literature has not measured an improvement in per-trade loss profiles for active participants under any documented regulatory regime"? The structural answer is that consumer protection in 2026 reduces participation but has not been demonstrated to change per-trade loss profiles.

  6. The slow-MM case at firm vs category level. Phoenix → Phoenix Legacy is a category-level case (the spot CLOB design lost the market). Wintermute's 90% liquidity drop on Hyperliquid is a firm-level case (a winner becomes a temporary loser). Should the chapter develop the slow-MM loser category at the firm or the category level? My current recommendation is category — because the named firm-level events are about volatility within the winner set, not about the category of slow MMs being displaced.

  7. The Pump.fun cohort vs the structural Pump.fun population. The CoinGecko data measures Pump.fun's monthly active wallets and their monthly P&L. The chapter should be careful that the framing accurately distinguishes between (a) the platform's profitability statistics improving because losers leave and (b) the platform's chronic-loser exit being itself a documented loser-category event. Is the chapter authorised to make the structural claim that the Pump.fun population data is the cleanest single piece of evidence in the book that the structural loser category is real and quantifiable? My recommendation: yes.

  8. The chapter's relationship to the "is this bad?" question. The Book Bible's clinical guidance says "the honest answer is sometimes 'this is just market structure' and sometimes 'this is a transfer from passive LPs to informed traders'; say which." Ch 11's clinical core is that the structural mechanism is the same on all three chains — concentration produces categorical losers — and the consent profile differs across the four loser categories. Is the chapter authorised to frame this as "the structural shape is unambiguous; the moral answer depends on the consent profile of each loser category"? My recommendation: yes, but I want Nick's sign-off before drafting the verdict section.

  9. Forward-looking trends Ch 11 should not preview. Ch 12 covers intent-based architectures, appchains, and cross-chain execution. Does Ch 11 anchor any forward-looking observations (e.g., Vitalik's 1-ETH solo staker proposal; Alpenglow's stake threshold drop from 4,850 SOL to 450 SOL; Hyperliquid's validator set 24→27) or does it stop strictly at the May 2026 state and leave the forward-looking material entirely to Ch 12? My recommendation: anchor the structural shapes but explicitly defer the forward-looking analysis to Ch 12.

7. Suggested structure for Phase 2 OUTLINE ​

The chapter is the first chapter of Part IV ("The Verdict"). It is not the book's verdict — it is the structural verdict, with Ch 12's forward-looking framing landing on top.

A workable outline shape:

  • Cold open. The Pump.fun cohort. 30.1% → 73.3% profitable wallets, May 2025 to April 2026; 5.2M → 1.8M monthly active wallets; the improvement reflects exodus, not conversion. The chapter's structural anchor.
  • What this chapter answers. Four questions corresponding to the four loser categories, plus the "why don't they leave" structural question.
  • The setup. The four categories defined. The consent-profile distinction. The three architectures (Ch 8 / 9 / 10) as the chapter's coordinate system.
  • The mechanics, in detail. Four subsections — one per loser category. Retail traders unaware of toxic flow (anchored on Pump.fun + BIS + Banana Gun); passive LPs in informed-flow pools (anchored on LVR + Topaze + Uniswap Labs); slow market makers (anchored on Phoenix + Wu et al. + the Solana access-vs-operational decomposition); validators without infrastructure relationships (anchored on the 770 → ~620 Solana exit + Hyperliquid 525K HYPE threshold + Lido 24% / solo staker 5.4%).
  • Why each group continues to participate. The structural "leave the table" subsection. One paragraph per category. The consent-gap framing.
  • How this plays out on each chain. The standard three-chain comparison. On Solana, the four losers are predominantly the unaware retail trader + the SFDP-removed validator; on Hyperliquid, the four losers are predominantly the HLP depositor in a JELLY-style event + the slot-#22 validator; on Ethereum, the four losers are predominantly the Banana-Gun user + the LP in V3 ETH/USDC 5-bps + the solo home staker (winner-by-yield, loser-by-share). The pattern: same four categories, different per-chain instantiation.
  • Who wins, who loses, why. The verdict. Clinical, not moralistic. The structural mechanism (concentration produces categorical losers) is the same across all three chains. The consent profile differs. The chapter must say which.
  • What changes when… Transition to Ch 12. Three forward-looking levers (intent-based architectures; permissionless top-N expansion; cross-chain execution) might or might not change the structural shape. Ch 12 develops which.
  • Footnotes and sources. Targeted at ~20–25 numbered citations. Heavy reliance on cross-references to Chs 2, 3, 7, 8, 9, 10 (the new citations are limited to the retail-loss + Pump.fun + Banana Gun + BIS + FCA + DEX-CEX-ratio anchors).

Word target: 3,500–4,500 words. The chapter is structurally complex but should be one of the shorter chapters in the book — the heavy lifting has been done in Chs 2–10; Ch 11's job is to land the categorical loser argument, not to re-litigate the per-chapter empirical work.

8. Sources cited (deduplicated) ​

  1. CoinGecko — Pump.fun Traders Are Making a Comeback
  2. CoinInsider — Pump.fun 73% profitability April 2026
  3. Phemex — CoinGecko reports 73.3% Pump.fun profitable
  4. BIS Bulletin No 69 — Crypto shocks and retail losses
  5. Investment Executive — BIS retail loss study
  6. Ecofin Agency — BIS 47.89% average retail loss
  7. FCA — Cryptoassets Consumer Research 2025
  8. FCA Wave 6 PDF
  9. CoinDesk — UK crypto ownership decline 2025
  10. CryptoSlate — Topaze Blue Uniswap V3 LP study
  11. Milionis, Moallemi, Roughgarden, Zhang — LVR (arXiv:2208.06046)
  12. a16z crypto — LVR explainer
  13. CoW — What is LVR
  14. Uniswap Labs Research — When V3 Returns More Fees for Passive LPs
  15. CrocSwap — Discrimination of Toxic Flow in V3 Part 4
  16. Sandmark — Toxic Flow: The Hidden Cost
  17. KuCoin — HLP Vault explainer
  18. ARX — Hyperliquid Vaults Explained 2026
  19. DefiLlama — Phoenix protocol page
  20. Ellipsis Labs — Introducing Phoenix Perpetuals
  21. Wu et al. — arXiv 2405.01329 (Ethereum builder market)
  22. Pahari & Canidio — arXiv 2509.16052 (exclusive transactions)
  23. TechFlow — BTC Liquidity 90% drop (Wintermute Hyperliquid)
  24. Syndica — March 2026 Solana onchain activity deep dive
  25. Chorus One — Timing Games on Solana
  26. Placeholder VC — Leveling the Stakes on Solana
  27. CCN — Solana 70% validator decline
  28. Phemex — Solana SFDP May 2026 requirements
  29. Blockworks — Solana Foundation pruning validators
  30. Helius — Solana Foundation Delegation Program
  31. Hyperliquid Docs — Staking
  32. cp0x on X — slot 21 ~525K HYPE
  33. Coin Edition — Hyperliquid 24→27 validators
  34. CoinDesk — Lido share August 2025
  35. Datawallet — Ethereum Staking Statistics 2026
  36. AInvest — Ethereum Staking solo vs Lido 2026
  37. Cointelegraph — Vitalik solo staking 1 ETH advocacy
  38. GitHub — ethstaker solo-stakers dataset
  39. CoinGecko — Top Telegram Trading Bots
  40. BlockBase — Banana Gun Research
  41. Thoma — Prospect Theory and Crypto Returns (SSRN)
  42. AInvest — Meme Coin Volatility psychological drivers
  43. CCN — Crypto Users Shifting CEX to DEX
  44. CoinGecko — DEX to CEX Volume Ratios
  45. ESMA — PFOF exemption list
  46. Proskauer — SEC withdraws 14 rule proposals
  47. Sumsub — MiCA Guide 2026
  48. ESMA — MiCA activities page

Cross-references to prior chapter RESEARCH.md and DRAFT.md files (Chs 2, 3, 5, 6, 7, 8, 9, 10) are used throughout to avoid re-litigating sourced material.