Gap Analysis

Available in: Builder

Gap Analysis analyzes your existing repertoire to identify potentially missing moves based on how frequently they occur in real games. Using probabilistic models from Lc0 or Maia, the system compares your current lines against expected human play, highlighting gaps where important continuations are not yet covered. This provides a fast, practical way to assess the completeness of your repertoire and systematically address weaknesses.

Clicking Run Analysis processes your repertoire using the selected engine and settings. Once complete, the results are displayed in a paginated table for easy review and action.

This feature helps you:

• + Analyze your repertoire using probabilistic models trained on millions of real games
• + Identify commonly played moves that are missing from your existing lines
• + Instantly jump to, add, or dismiss gaps directly from the analysis results
• + Improve the practical completeness and reliability of your opening preparation

Gap Analysis

Once the analysis has been run, the results are presented in a paginated table showing all detected gaps that meet the configured criteria.

For each identified gap, this section displays:

• ◦ Move: The missing move identified by the selected engine, representing a common or highly probable continuation not currently present in your repertoire.
• ◦ Eval / Probability: If using Lc0, this shows the engine evaluation for the move. If using Maia, this shows the probability of the move being played by humans at the selected rating level.
• ◦ Ply: The ply number at which the missing move occurs within the game tree.
• ◦ Actions: Provides quick controls to Go To the position, Add the move directly to your repertoire, or Dismiss the gap if you choose not to include it.

At the top of the table, the total number of detected gaps is shown along with the active probability threshold, which dynamically filters the list to include only moves meeting or exceeding the configured minimum likelihood.

Note: You can rerun the Gap Analysis at any time to reflect changes in your repertoire or settings. Existing results can also be deleted, allowing you to start a fresh analysis whenever needed.

Gap Analysis Settings

The Gap Analysis can be configured using the following options:

• ◦ Engine: Choose between Lc0 or Maia. When Maia is selected, you can specify a target rating level between 1100 and 1900 to match expected human play.
• ◦ Minimum Threshold Percentage: Filters out low-probability moves by defining the minimum likelihood required for a move to be considered a gap.
• ◦ Ply Range: Controls which portions of the game tree are scanned by defining the minimum and maximum ply to analyze.
• ◦ Find Gaps: Limits the analysis to missing moves for White, Black, or Both, based on how your repertoire is structured.

Note: Lc0 and Maia identify gaps based on the probability of moves being played rather than strict engine optimality. Because these models are trained on millions of games, their predictions are highly stable and reflect realistic human expectations rather than theoretical perfection.

If you want to know more about how these models work in practice, check out the following articles: Integrating Maia2 — the human-like chess engine — into my chess application and Using Maia and Leela Chess Zero to find repertoire gaps, which go deeper into Maia integration and gap-analysis workflows.