botunnet wrote: Thu May 01, 2025 1:55 pm
With the Unique openings, the draw rate is over
92%
only 372 games (=8% of 4680 games) are useful for separating engines

we just learn which engines blunder the most due to
lack of time in drawish positions

it's only an
useless lottery.
With the UHO openings, the draw rate is barely
56%
2055 games (=44% of 4680 games) are useful for separating engines

we learn which engine knows how to capitalize on an
advantage, which engine resists under
pressure on non-drawish positions, which openings should be
trained, etc.
To know which engine is the strongest in general, we have to cover as many ECO codes as possible and avoid non-discriminating positions (like the starting position, drawish opening lists, heavy unbalanced positions, positions with poor mobility of pieces, etc.)

Unique openings are not relevant.
To improve the engine's evaluation, we needn't tons of moves evaluated at 0.00 regardless of depth.

Unique openings are not relevant.
At SedatChess, tourney's conditions only favor blunders due to shallow depths due to a lack of time/power. This is pointless because simply by giving more power or time, these blunders disappear so there would be more draws and we already know that the Unique openings produce a draw rate over 90%.