What are the odds a 2800 GM can beat SF 8 in one game?

[jplchess]

Computer can do 3300 ELO and game in 3 hours

Jonathan Lee

[Milos]

If Stockfish is playing on modern cellphone 1 in 100. If Stickfish is playing on TCEC hardware 1 in million.

[Jhoravi]

If the GM plays the English opening Botvinnik variation as while and Stockfish allows white to play f5 then yes.

[IWB]

Computer can do 3300 ELO and game in 3 hours

Jonathan Lee

Time and Computer dosen't mater if you have a defined Elo.

2800 to 3300, or 500 Elo difference, is a "winning" ratio of 5.32% according to ELo formula. That means that the 2800 get 5.5 Points out of 100 games. If that are 5 wins and one draw or 11 draws is uncertain. And if the result is higer or lower than 5.32% the difference was bigger or smaller than 500 Elo. Which means you can tell the rating after a game, not before.

Your problem is the 3300! Where do you get that number from? How do you know that SF8 has 3300? Whoever comes up with a rating of 3300 for SF8 made an eduacated guess - at best!

Ingo

[Milos]

2800 to 3300, or 500 Elo difference, is a "winning" ratio of 5.32% according to ELo formula. That means that the 2800 get 5.5 Points out of 100 games. If that are 5 wins and one draw or 11 draws is uncertain.

It's very certain. Try and you'll see, at least 9 out of 10 times it will be 11 draws (or more correctly no wins for weaker engine).
Plus 5.32% rate for weaker engine is not according to "Elo formula", but according to the model used. Different model different result.

[IWB]

2800 to 3300, or 500 Elo difference, is a "winning" ratio of 5.32% according to ELo formula. That means that the 2800 get 5.5 Points out of 100 games. If that are 5 wins and one draw or 11 draws is uncertain.

It's very certain. Try and you'll see, at least 9 out of 10 times it will be 11 draws (or more correctly no wins for weaker engine).
Plus 5.32% rate for weaker engine is not according to "Elo formula", but according to the model used. Different model different result.

different model, different result ... maybe, but I am sure he asked for this model: https://en.wikipedia.org/wiki/Elo_rating_system and there 500 Elo difference is 5.32%!
9 out of 10 is 'very certain' for you ... thats funny

[Milos]

different model, different result ... maybe, but I am sure he asked for this model: https://en.wikipedia.org/wiki/Elo_rating_system and there 500 Elo difference is 5.32%!

Even though you are quoting wiki, it seams you never actually read that page, otherwise you'd know this: https://en.wikipedia.org/wiki/Elo_ratin ... cal_issues
You could also try reading about tails of distribution of the model used and how accurate those are, but why bother, 500Elo = 5.32% is like written in stone, right?

9 out of 10 is 'very certain' for you ... thats funny

I see, for you whatever you personally didn't try has 50:50 chance.
Chance of 5 wins for 500 Elo lower engine that you quoted is less than 1 in million. Well for you that is uncertain.
So you probably have 50% chance beating Carlsen since you haven't tried it, and outcome is uncertain, right?

[h1a8]

Odds are 0

[reflectionofpower]

Odds are 0

Agreed

[Lyudmil Tsvetkov]

so, if the odds are one to a million, and I put 1 buck on a win in a 10-game match against SF, I will be a millionaire after the match.

certainly the easiest way to get rich...