Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Enhancement: Use ΔScore in calculation of top 3 moves for handicap games #1343

Closed
benjaminpjones opened this issue Jan 3, 2021 · 2 comments · Fixed by #1553
Closed

Enhancement: Use ΔScore in calculation of top 3 moves for handicap games #1343

benjaminpjones opened this issue Jan 3, 2021 · 2 comments · Fixed by #1553

Comments

@benjaminpjones
Copy link
Contributor

I believe score is a better indicator of major events in the game, but this is especially prominent in handicap games. See discussion at Top 3 Moves: Score a Better Metric?. I assume this would be a backend change, but if I'm wrong, please let me know!

Proposed Solution
For handicap games, instead of the top changes in win rate, display the top changes in score. I think using win rate still has merit in even games (particularly high level ones).

@anoek
Copy link
Member

anoek commented Jan 3, 2021

I think this could be prototyped on the client for supporter games that are fully reviewed, I believe we recalculate those top 3 moves there since the top 3 moves are based off of a quick scan of the game not the fully reviewed data. But yeah once we get it dialed in to work the best it can work, I'll need to port that over to the server side to work for non supporter games.

If someone does this, keep in mind that score isn't available for Leela Zero reviewed games, only Kata Go games (although katago is the default and by far most used at this time)

@benjaminpjones
Copy link
Contributor Author

Thanks for the additional info!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants