docs: 100bb chart validation v2 — limp-collapsed apples-to-apples#169
Closed
amaster97 wants to merge 1 commit into
Closed
docs: 100bb chart validation v2 — limp-collapsed apples-to-apples#169amaster97 wants to merge 1 commit into
amaster97 wants to merge 1 commit into
Conversation
Re-run of 100bb HU preflop validation comparing engine output to the published chart under TWO views, so View B is structurally comparable to the chart's fold/raise-only categorization at SB root: - View A (as-solved, raw 4 actions): [fold, call(limp), open_2.5bb, all_in] - View B (limp collapsed into raise at SB root): chart's no-limp convention View A SB-RFI: 15.67% fold + 47.05% limp + 37.28% open = chart-incomparable. View B SB-RFI: 15.67% fold + 84.33% raise+ vs chart's 21.65% fold + 78.35% raise → aggregate gap collapses 41pp → 5.98pp. Per-cell match: View A 26/37 (70.3%), View B 28/37 (75.7%). 98s and 22 flip NO→YES under View B (both 100% limp in View A; under no-limp convention both become raise). Remaining 9 cells diverge for non-limp reasons: BB-vs-RFI A5s/A2s suited-ace flat vs chart's bluff-3bet (Nash multiplicity), KQs over-3bets at 8.5bb vs chart's call at 10bb (sizing mismatch), JTs/76s/J7o/Q5o defensible-flat-call vs chart's bluff-or-fold (range-cap convention). 2500 iters, 88.4s wall (vs V1 10k @ 348s). Premium cells unchanged (AA, AKs, all pairs JJ+, all 3-bet/4-bet jam patterns). Cleared-by-View-B aggregate is the load-bearing finding — the 41pp SB-RFI gap in V1 was a labeling artifact, not a solver bug. Engine code untouched; .so reused from origin/main (PR #165 + PR #167 build). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Owner
Author
|
Closing as duplicate — a parallel agent pushed the same v2 validation doc directly to origin/main (commit 9204192) reaching the same conclusion (View B 75.7% match, 88s wall, same per-cell results). Both reports concur. Closing this PR; the merged doc is already at docs/preflop_100bb_chart_validation_v2_2026-05-28.md on main. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
[fold, call(limp), open, all_in]but published chart prescribes{fold, raise}only).Findings
98s and 22 flip NO→YES under View B (both 100% limp in View A, become raise under no-limp convention). Remaining 9 mismatches are non-limp:
Verdict: PARTIAL. View B confirms V1's largest gap was a labeling artifact, not a solver bug.
Test plan
reraise_multipliers=[5.0]to emit 10bb 3-bet and isolate sizing-mismatch contributionsDocs-only PR. Auto-merge OK per task brief.
V1 doc | V2 doc
🤖 Generated with Claude Code