How our analysis works

Separate the numbers from the strategy

Pot odds and required equity

A chip-based threshold calculated from the additional call and the pot you are eligible to win. It is not your chance of beating an opponent. Contributions are considered by player; side pots you cannot contest are excluded.

Equity against a range

Opponent hand candidates are inferred from public actions. Normally, candidates and boards are sampled. On the river, with one opponent, no further decision after calling and verifiable pot-allocation conditions, all candidate hands are enumerated. The calculation does not inspect the opponent’s actual hidden hand or a future board. Even full enumeration does not remove uncertainty in the assumed range.

Educational baseline strategy

A reference based on position, stacks, betting history, board and other conditions. The current game coach is not an exact GTO solver. Multiple candidates do not imply equal EV or a particular mixed frequency. For a short stack outside the blinds facing a small open, reference candidates are shown without grading solely by deviation from them. Unverified situations are also excluded from scored questions.

Adjustments to opponent tendencies

Practical adjustments based on assumptions such as tight or loose play. An opponent entering many pots does not necessarily call large bets with weak hands. Explanations distinguish these assumptions from the baseline.

ICM evaluation

A model for prize equity based on payouts and each player’s stack. Where limited conditions permit calculation, a separate break-even calling threshold is shown. Complex multiway spots or situations with further betting are not presented as exactly solved values.

“Above the required equity” is not enough by itself

If the pot including the opponent’s bet is 30 BB and calling costs another 10 BB, the chip-based threshold for a single pot is:

10 ÷ (30 + 10) = 25%

That 25% applies when you can reach showdown without paying more. A player behind who can raise, future postflop betting or different side-pot eligibility makes a simple comparison insufficient. In a multiway pot, multiplying separate heads-up win probabilities is not a valid shortcut.

When the conditions needed for evaluation are not met, the coach withholds a recommendation or score. “Cannot evaluate” should not be recorded as a mistake.

Using WHY Coach

Start with the conclusion and main reason. Open “Other options,” “If conditions change” and “Calculation assumptions” when needed. Aim to take away one principle that still helps when ranges or bet sizes change.

Explanations are assembled from the current situation and calculation results. Natural-language follow-up chat is not currently available.

Leak Finder and your next practice

Member records from lessons, challenges and AI games suggest themes to review and related practice. Difficulty and data types are separated, so repeating one question does not inflate apparent ability.

Choose a 30-, 90- or 365-day analysis window. Successfully saved AI decisions accumulate across tournaments for members. The analysis covers the last 365 days, up to 10,000 records; older records are trimmed when saving. Course weakness summaries use first answers. Challenges use the latest 60 sessions within the selected window. Unsaved decisions from past tournaments are not reconstructed.

Reattempted course questions are saved as review records, separate from initial results. Comparisons require verifiable question content and the original evaluation version. Missing past reviews are not reconstructed, and answering the same question correctly is not treated as proof of transfer to new situations.

Focused practice combines mistakes to review with related questions you have not previously answered in this service, drawn from the same unlocked course’s lessons and a dedicated focused-practice bank. Dedicated questions include both calculations from stated amounts and reasoning under specified opponent tendencies or spot conditions. Their answers do not guarantee a universally GTO-optimal action. Dedicated answers do not change initial course results, XP or rank. Changing suits alone does not create a new question. If too few related questions are available, the interface says so. A first correct answer is a learning signal, not proof of real-game mastery from a small sample.

Historical comparisons separately compare the latest 14 days with the preceding 14 days for matching question type, difficulty and evaluation version, regardless of the selected analysis window. Small samples and older evaluation standards are not treated as conclusive weaknesses. These trends are learning signals, not measured EV lost in real play. Suggested practice still follows course-unlock requirements and membership usage limits.

Spacing answers to plan the next review

Saved question content, evaluation version and answer time determine review timing. The initial interval is 24 hours; correct answers after the scheduled interval move the target to 3, 7, 14 and 30 days. A mistake returns it to 24 hours. An early correct answer does not advance the stage. Timing uses elapsed hours, not calendar-day boundaries.

Growth records distinguish reviews within 24 hours, reanswers at least 24 hours later and first answers to related questions. The latest answer per question in each category is shown, rather than treating repeat counts as ability. Records whose earlier content or time cannot be verified are excluded from spacing comparisons.

Within the same theme and unlocked course, questions prioritize due reviews and prior mistakes. Recent mistakes over 30 days adjust the sequence of new related calculation and condition-based questions. Review schedules always use the last 365 days, up to 10,000 saved records, independently of the analysis-window selector.

This is a continuing-study policy, not a prediction of individual memory or real-game mastery. Times when explanations were viewed on another screen are not measured. Schedules are recalculated when the screen opens or refreshes and show the check time. Comparisons are recalculated when learning materials or records are insufficient.

High-precision GTO analysis: current limits

Solver-derived EV and recommended frequencies are not currently provided.

Preparation focuses on validating legitimately usable solver output and applying it only to matching player counts, stacks, ranges, betting histories and payout conditions. AI-estimated EV or frequencies are not presented as solver calculations.

• Verified preflop analysis data: not yet integrated.
• Exact EV comparisons across all actions and bet sizes: not supported.
• General postflop, multiway and node-lock solving: not supported.

Deep GTO and PRO availability will be announced only after the accuracy and usage conditions of analysis data are confirmed.