OllinStudy

Methodology

What is solved, how exactly, and how you can trust it.

Regime

Heads-up MTT, 10–100bb effective. Every published strategy comes from a full-tree solve of the exact spot — no interpolation between depths, no preflop abstraction shortcuts, no borrowed 6-max ranges.

Accuracy

Solves are run to a Nash-distance of 0.3% of the pot (exploitability bound per player). Convergence numbers ship with each dataset and are re-verified by an independent gate before publication.

Boards

Postflop strategy is solved on 184 canonical flops — the strategically-distinct subset after suit isomorphism — and mapped back to all 1,755 strategically-equivalent flops at lookup time.

Dataset versioning

Every number in the product is traceable to a dataset stamp. Current dataset: 20260721T131354Z. When a new dataset is published, the stamp changes everywhere at once — mixed-dataset views are impossible by construction.

This page is a stub of the full methodology document; solver verification artifacts live in the solver-lab workspace and will be linked here when published.