Review a downswing in three separate passes

Pass 1: reconcile the record

1. Include every session in the period, including losses. Check deleted/duplicate or missing imported records against your own source.
2. Recheck buy-ins, rebuys, cash-outs, currency and duration. Keep bankroll transfers separate from playing profit.
3. Use the same cost policy throughout; do not subtract a fee twice when it is already reflected in cash-out or entry cost.

The combined rate is $60 ÷ 6 = $10/hour, not the simple average of +$30 and −$30. If a missing −$120 record is found later, the original conclusion was based on the wrong ledger. Verify totals before interpreting a chart.

Pass 2: ask whether the sample is comparable

Separate Cash and MTT, currencies, stakes, games and materially different conditions. Check whether the apparent slump coincides with moving to tougher games, longer sessions or a new format. Single-dimension filters are Free; combined filters and deeper analysis may require Pro. A convenient filter is not evidence that the remaining sample is representative.

Adjacent sessions may share fatigue, opponents or conditions. As NIST: consequences of non-randomness explains, related observations do not necessarily provide independent information. Applying that warning here is a reason to inspect your sample, not a calculation of poker skill.

Pass 3: review decisions without using the result as the verdict

Choose hands and notes for concrete questions: what information was available, which action was taken and what alternative deserves review? A winning hand can contain a mistake; a losing session can contain well-reasoned decisions. Record important spots in the hand recorder; the Session Tracker does not analyze hand decisions for you.

Write down a review question before studying the outcome. Compare the type of game you intended to play with the actual stakes and conditions in your records. This workflow is not a recommendation to move up, chase losses or continue playing.

Where the historical comparison fits

The Variance Lab guide explains resampling recorded cash sessions. NIST: bootstrap resampling describes drawing repeatedly with replacement to study a statistic's uncertainty. The result depends on the data and model; repeatedly sampling an incomplete history cannot supply missing kinds of results.

PokerArc's eight-session minimum is a software gate, not a reliability certificate. Its historical percentile is not a probability that you are unlucky, profitable or playing badly. Review MTT denominators separately, and reconcile shared Home Game entries before copying a personal result.