📚 Series overview — review page. This page renders the 4-article arc Dan's Squid Foundations series will cover, the thesis arc across articles, and how the series maps to Dan's original proposal. Individual articles each get their own outline + preview page (Article 1 is at /articles/dan-squid-mental-models/).

Status (2026-04-25): Series structure agreed (4 articles); production sequence locked (1 → 2 → 3 → 4); audience = casual + intermediate; data approach = theoretical-first; distribution = QuintAce + Dan's platform. Article 1 outline + Phase A solver pulls complete. Articles 2–4 sketched at thesis level, awaiting refinement after Article 1 ships.

Daniel Dvoress co-founded GTO Lab. This series teaches Squid Classic from first principles — not as "NLHE with wider ranges," but as a structurally new game where one rule inserts a new EV term into every decision. Inquiry-led prose throughout.

Series thesis

Squid Classic is 6-max NLHE plus one rule: at the end of each game, the last player without a "squid" pays a penalty to everyone else. That single rule inserts an invisible per-hand term — squid equity — into every decision. Once you can name that term, six things that look like "Squid is just different" stop being mysterious and start being computable.

This series teaches the mental model in four passes, each building on the last:

  1. Mental models — name the invisible ante; show the EV decomposition; build the napkin heuristic.
  2. Desperation geometry — when only two players are still desperate (no squid), the ante swallows the chip-EV calculation. Watch it happen.
  3. Future game EV / path of the squid — track one squid through an entire game and see why "winning hand 1" is not the same as "winning hand 6."
  4. Shallow vs deep — at low M the math bends back. The mental model needs a depth-axis.

A future companion series (Squid Exploits, separate, deferred) will cover the population-vs-optimal dimension that Dan's original plan included. Out of scope here.

How the 4 articles map to Dan's original proposal

Dan's original proposal had two top-level sections:

§1 — General differences from cEV (4 topics): 1. The cost of stack survival 2. The value of winning a pot at all 3. (experimental) Optimal play vs how population plays — potential exploits 4. Shallow vs deep regime breaks

Specific game states (3 chapters): - Ch 1 — Initial state: equity distribution / EV calculation at start of round - Ch 2 — Desperation: down-to-two scenarios (BvB, BTN vs SB, CO vs BTN) - Ch 3 — Game flow: path of the squid, future-EV across hand sequence

The 4-article restructure folds these as follows:

Dan's original Where it lands in the series
§1 topic 1 — cost of stack survival Article 1 §4a (Difference A: Desperate ranges widen)
§1 topic 2 — value of winning a pot Article 1 §4b (Difference B: Safe and desperate play different games)
§1 topic 3 — optimal vs population Deferred to "Squid Exploits" companion series (different shape; needs HUD/population data Squid doesn't yet have)
§1 topic 4 — shallow vs deep Article 4 (becomes its own article — the depth-axis is structurally important enough to merit standalone treatment)
Ch 1 — Initial state Article 1 §0–§3 (the foundational EV decomposition + naming squid equity)
Ch 2 — Desperation Article 2 (becomes its own article — three pairings × stack depths × val settings is its own data story)
Ch 3 — Game flow Article 3 (becomes its own article — sankey/flow viz + per-hand bb/100 across a game is a major data story)

Why the restructure: - Dan's §1 was a flat list; we layered it into a thesis arc (name the term in Article 1, watch geometry collapse in Article 2, watch flow play out in Article 3, watch depth bend it in Article 4). - Ch 2 (desperation) needs its own article because the 3-pairing × stack-depth grid is too dense to fit inside a "general differences" piece. - Ch 3 (game flow) needs its own article because hand-conditional simulation data is a major scope decision (could need a custom batch run). - §1 topic 3 (population exploits) is structurally different — it needs HUD data Squid doesn't yet have, and the framing is "where do real players deviate" rather than "how does the format change EV math." Best as a follow-up series.

Article 1 — "The Ante You Can't See: Mental Models for Squid Value"

Slug: dan-squid-mental-models · Outline + Phase A details → Length: 2,400 words · Type: plot-twist 🔀 · Status: outline + Phase A pulls complete; Phase B drafting unblocked Maps to Dan's original: Ch 1 (Initial state) + §1 topics 1–2 (folded into §4a/§4b)

Thesis: Squid Classic adds one mechanic — the squid penalty at game end — and that single change inserts an invisible per-hand ante into every decision. Once you can name that ante, three "differences from cEV" stop looking like separate rules and become the same machine viewed from three angles.

Structure (7 sections, ~2,400w): - §0 — Squid Classic in 60 seconds (200w) - §1 — Hook: a hand that costs more than it shows you (250w) — BTN J8s vs CO 2.5x open at val=3, cash 96.6% fold → Squid 0% fold - §2 — The cEV machine you already own (300w) - §3 — Naming the missing term: squid equity (450w) - §4 — Three "differences from cEV" that are really one mechanic (700w): - §4a — Difference A: Desperate ranges widen (Dan's original §1 topic 1) - §4b — Difference B: Safe and desperate play different games (Dan's original §1 topic 2) - §4c — Difference C: Bluff frequencies drop because defenders overdefend MDF — CO c-bet on 765two wet flop, cash 69.5% → Squid 55.5% (-14 pts) - §5 — Napkin math: spot the squid equity (400w) - §6 — Closer (200w)

Widgets: 9 widgets across 7 sections; 5 anchored to PokerQuiz inventory shapes (PayoffPicker, EvCalc, BubblePressure, ZeroSumTug, BackwardSolver).

Data: Theoretical-first with fresh solver verification on hook + practice spots + bluff-magnitude. 12/12 fresh queries successful (Phase A complete 2026-04-25).

Article 2 — "Down to Two: Desperation Geometry"

Slug: dan-squid-desperation-geometry · [outline TBD] Length target: 2,400 words · Type: mechanism walkthrough 🔬 · Status: thesis-locked, refinement after Article 1 ships Maps to Dan's original: Ch 2 — Desperation (down-to-two scenarios)

Thesis (working): When a Squid Classic game narrows to two desperate players still without a squid, the geometry collapses. Three pairings (BvB, BTN vs SB, CO vs BTN) each rewire differently because the ante size is no longer abstract — it's about to fire on whoever loses the next significant pot. Watch the chip-EV math get swallowed in real time.

Tentative structure: - Three pairings, three sub-sections (BvB · BTN vs SB · CO vs BTN) - Each pairing: theoretical setup, solver pulls, contrast with cash equilibrium - Closer: the "geometry-collapse" pattern that emerges across all three

Data approach: Measured. Real solver pulls per pairing, per stack depth. Estimated 8–12 pulls. Major scope question: does the rail extend to "two players left desperate" game state, or only to "hero is last desperate" (the existing SquidConfig.hero_last)?

Widgets: 4–5 (pairing selector · chip-EV-vs-total-EV split bar · spot card per pairing · "what would you do" quiz)

Open editorial questions: - Q: Does book-2 v1.8.0 explicitly cover all three pairings, or only some? - Q: Is "two desperates left" a solver-supported game state, or do we need to construct it via end-of-game proxy?

Article 3 — "The Path of the Squid: Future Game EV"

Slug: dan-squid-future-game-ev · [outline TBD] Length target: 2,600 words · Type: plot-twist 🔀 · Status: thesis-locked, major data scope decision pending Maps to Dan's original: Ch 3 — Game flow (future-EV across hand sequence)

Thesis (working): Track one squid through an entire 6-handed game. Hand 1's winner gets the squid; that single transfer reshapes Hands 2–6 for everyone at the table. We measure the bb/100 ripple — which seats benefit, which seats now play scared, and why "winning Hand 1" is the most under-appreciated equity term in Squid.

Tentative structure: - §0–§1: setup — pick a 6-handed game, identify the squid-acquisition events - §2–§4: hand-by-hand walkthrough — what changes after Hand N's squid lands? - §5: aggregate bb/100 ripple per seat - §6: closer — the value of winning Hand 1 in BB equivalents

Data approach: Measured. Sankey-style flow data — hand-by-hand bb/100 conditional on game-state. Could be 15–20 pulls or could be derived from a single batched simulation. Major scope decision: does book-2 v1.8.0 surface hand-conditional solver data, or does this article need a custom multi-hand batch run upstream?

Widgets: 4–6 (sankey/flow diagram · "where does the squid go" picker · per-seat bb/100 timeline · spot-card sequence for hands 2/4/6 · summary "value of winning Hand 1" calculator)

Open editorial questions: - Q: Can we get hand-conditional solver data (e.g., "Hand 4 strategy given hero already has the squid")? If not, this article needs a custom batched simulation. - Q: Distribution of squid-acquisition timing — does book-2 measure this, or do we estimate from first principles?

Article 4 — "Shallow vs Deep: Where the Mental Model Bends"

Slug: dan-squid-shallow-vs-deep · [outline TBD] Length target: 2,200 words · Type: myth-busting 🧨 · Status: thesis-locked, refinement after Article 3 ships Maps to Dan's original: §1 topic 4 — shallow vs deep regime breaks

Thesis (working): The ante framing from Article 1 assumes deep stacks. As effective stacks compress, the squid's relative size grows — until at low M the chip-EV term swallows the squid term, and the mental model inverts. We map the depth-axis: where the ante dominates, where it's a tax, and where it disappears.

Tentative structure: - §0: depth as the missing axis - §1–§3: cross-depth pulls on 2–3 anchor spots (effective stacks 8bb / 25bb / 50bb / 100bb) - §4: the regime taxonomy — three depth zones, what happens in each - §5: napkin math for depth — when to switch mental models - §6: closer + connection to book-2's M-Probe / M7 work (if landed)

Data approach: Measured. Cross-depth solver pulls on 2–3 anchor spots. Estimated 10 pulls.

Widgets: 3–4 (depth slider on a single anchor spot · cross-depth comparison grid · "where would you stop" quiz · summary depth-regime cards)

Open editorial questions: - Q: Is book-2's M-Probe / M7 graded data ready as of v1.8.0, or still WIP? If ready, this article cross-references; if not, this article stays self-contained on depth axis only.

Anti-overlap matrix (vs 4 already-shipped Squid articles)

This series is the foundational mental-model arc. It must NOT overlap with what's already shipped:

Already shipped What it covers Why this series doesn't repeat it
nick-squid-vs-nlhe-first-orbit On-ramp for cash regs; reflexes that need swapping in first orbit Nick's piece is the elevator pitch for someone sitting down at their first Squid table. Dan's series is the deep mental model the on-ramp gestures at.
nick-desperation-polarization Hero-last preflop cliff (BB last to act after folds) + K94ss postflop flip Nick's "hero-last" = one player's solver state on a single hand. Dan's "down to two" (Article 2) = game-state when only 2 players remain desperate. Different scenarios, complementary.
dan-cbet-wrong-in-squid Dan's own postflop c-bet article on three textures Postflop sizing tree, narrow scope. This series goes upstream — why the c-bet tree breaks (the ante) and downstream — how the ante shapes the whole game.
uri-nlhe-reflexes-to-unlearn (WIP) 7 NLHE habits that leak in Squid, with replacement rules Tactical / habit-level. This series is theoretical-foundational — it's the math underneath Uri's habits.

Production sequence

Strict 1 → 2 → 3 → 4 order. Each article cites the previous.

Article Dependencies Drafting unblocks
1 None (theoretical) All — establishes named ante used by 2/3/4
2 Article 1's named ante 3 (game-flow needs pairing terminology)
3 Articles 1–2; major data scope decision 4 (depth-axis builds on path-of-squid)
4 Articles 1–3; book-2 M-Probe alignment Series complete

Coach checkpoint cadence: Single async checkpoint per article (Dan reviews live URL post-deploy). No mid-draft coach gating; voice corpus carries it.

Internal review cadence: Tier 2b 9-agent review on Articles 1, 2, 3 before deploy. Article 4 may go Tier 2a (lighter) since it builds on three vetted pieces.

Distribution plan

Primary: internals.quintace.ai (existing rail; same as Dan's c-bet article) Secondary: Dan's own platform / GTO Lab (if Dan opts in; format TBD) Cross-promotion: each article ends with a 1-sentence forward link to the next; Article 4 closes with a link back to this series-overview page.

Cadence question: batch all 4 and drop together, or stagger weekly? Stagger lets Dan react to Article N's reception and tune Article N+1. Recommend stagger weekly.

Cross-series open questions (resolve before refining individual articles)

  1. Hero-secondlast solver coverage — Article 2 (down to two) needs solver data for 2-player-desperate game state. Does the existing rail (strategy_grid_client.py / SquidConfig.hero_last) extend to "two players left desperate" or is that a gap? Scout before committing Article 2 data plan.
  2. Hand-conditional batched data — Article 3 needs hand-by-hand game simulation. Does book-2 surface this, or does it require a custom multi-hand batch upstream? Major scope decision; resolve before Article 2 ships.
  3. M-Probe alignment — Article 4 cross-references book-2's depth-regime work. Confirm book-2 v1.8.0 actually contains M-Probe / M7 graded data Dan can reference. Verify before Article 4 drafting.
  4. Series landing page — build a dedicated landing page on internals (this series-overview page becomes the canonical entry point), or just inter-article links? This page IS the v1 of the landing page.

Status log