7-piece Syzygy tablebases are complete

🗝️🗝️🗝️🗝️🗝️

Does anybody know the distribution % of Win/Draw/Loss in this 7 man TB with 400 trillion positions ??

So something like: Win - 10%, Draw 75%, Loss 15% over all positions??

@cr2015 syzygy-tables.info/endgames has these stats for all individual endgames (including in machine readable format: syzygy-tables.info/stats).

You'd still have to add them up, but how do you add KPPvKQ to KQvKR? Lexicographically (KPP + KQ vs. KQ + KR)? Strongest by sum of material (KQ + KQ vs. KPP + KR)?

Well, one could simply add up draws vs non draws.
I would expect most positions to be not drawn since there are a lot of "unfair" material distributions.

Shouldn't a table base be perfectly symmetric since it will include every position with colors swapped. So the results will always be equal.

So if you can simply determine the draw percent, you should be able to derive the win percent. E.g. - if 70% of games are drawn then that would imply that 15% of the time white wins and 15% of the time black wins. I'm not sure these values mean much of anything though.

You're right. While the normalized tables are not symmetric (the side move move might be winning more often than the other side, or KQ may me winning more vs. KP than KP is winning against KQ), chess is of course symmetric. 13% of unique 7-piece endgames are draws, 0.3% draws because of the 50-move rule.

@revoof

the 13% draw is a real number? so 87% win/loss as per @OhNoMyPants comment ?

@cr2015 Yes, real number calculated from the tablebase stats.