Checker-play analysis compares legal alternatives
An engine evaluates candidate plays from the same position with the same dice. Its preferred move is the one with the highest estimated value under the specified rules, score and evaluation settings. It is not simply the move that hits most often or wins the greatest percentage of single games.
When reviewing a mistake, compare what the alternatives change: board strength, contact, timing, gammon exposure and the race. If the best play leaves a blot, look for the positional gain that compensates for the risk. “The engine says so” is an answer to which move, not an explanation of why.
Cube analysis includes decisions not to act
Checker errors happen after a roll; a missed double can happen before one. Cube review compares the value of not doubling with offering a double, including the opponent’s best response. When a double has been offered, take and pass are separate alternatives.
This distinction matters when studying a report. A recommended double does not mean the opponent should pass. Likewise, a missed double can be a serious error even if the checker play that followed was excellent.
Equity measures value, not certainty
Equity is the expected value assigned to a position or decision. Money-game equity is commonly expressed in points relative to a cube unit; match analysis may use match-winning chances or a normalised equity scale. Always check the unit before comparing values.
Equity loss is the difference between the estimated best alternative and the chosen action on the same scale. It is not the probability that a move will immediately lose. A small disadvantage repeated across many decisions can matter, while one unlucky reply does not prove the earlier choice was wrong.
Errors and blunders are review priorities
Labels such as error and blunder identify decisions with larger estimated losses. Their thresholds and presentation depend on the analysis system. Avoid comparing a label across different tools as though it guaranteed the same underlying loss.
Start with the largest losses, but group them by cause. Three missed doubles in similar attacking positions may teach more than three unrelated checker errors. Check whether you failed to see an alternative, misjudged a position or forgot the match score; each needs a different practice exercise.
Use BlitzGammon’s post-match review
BlitzGammon uses GNU Backgammon (GNUBG) for post-match analysis. Its match review includes player performance summaries, checker-play analysis and cube-decision information. Reviewing your decisions is separate from the match result: winning does not establish that your moves were more accurate.
Analysis availability and depth follow the existing account access rules. A report can take time to complete, and deeper analysis is not an instant or infallible verdict. This guide does not promise the same analysis depth for every account or position.
Study a single position in Position Analyzer
BlitzGammon’s Position Analyzer is a member-gated workspace for setting up positions and analysing checker plays or cube decisions. It supports XGID import as well as board editing. Enter the player on roll, dice when relevant, match context and cube settings carefully before asking for analysis.
A useful exercise is to recreate a decision, choose two candidate plays and write down your reasons before viewing the result. Then alter one feature, such as the cube owner or a home-board point, and predict whether your conclusion should change. Only compare results obtained with the intended context.
A repeatable review routine
Review a small number of important decisions rather than clicking through every mistake. First reconstruct your original reasoning. Next identify the engine’s alternative. Finally explain the difference in ordinary backgammon terms and choose a related position to practise.
Do not overinterpret tiny gaps. Engine settings and sampling can affect close comparisons, and an analysis result is an estimate rather than mathematical proof. The practical goal is to recognise recurring situations earlier in your next game.
- Record one checker-play lesson and one cube lesson.
- Distinguish a decision error from a surprising dice outcome.
- Revisit the lesson after several matches to see whether it recurs.