A 3000 could easily beat a 2000, but could a 4000 easily beat a 3000?

From 1931 to 1937, Vera Menchik participated in 4 World Championships, winning all 45 games. Not a single loss, and not a single draw. (2 of the 45 were by forfeit when her opponent could not travel to tournament).

She did lose to men in the meantime. And in matches against Sonja Graf.

What was the Elo of the rest of Vera´s opponents in women´s world championships?

If someone showed up and played Carlsen and the rest of the 2800+ Elo bunch with similar results - all wins, in 45 games - what would the result of Elo computation be for 45 wins but no losses and no draws?

Statistically, if a player won 100% of all games, his rating would be infinity.

However, FIDE elo calculator accept maximum of 800+ than his opponents.

So if he won 45/45 games vs 2800 players= 2800+800= 3600 rating!

If Elo were used in a pure zero-sum game, a 4000 would beat a 3000 equally as often as a 2000 would beat a 1000.

But chess has an upper limit due to the nature of the game being a theoretical draw. The highest possible chess rating is somewhat below 3600, according to chess experts and statisticians.

I would like to know how they got 3600 maximum cap, do you have a link?

I saw the same thread in talkchess in about 2 years ago, where majority of people agreed between 3800-4500. (As usual, I guess that idea came from statistic expert Kai Lasko but I could not find the link now.) 

The answer is yes.  Look at the last TCEC final, it was a slaughter.

mecuelgalapieza wrote:

The answer is yes.  Look at the last TCEC final, it was a slaughter.

TCEC Season 116 Jan – Apr 2018 Stockfish 260318 Houdini 6.03 + 20 = 78 - 2

Do you mean a slaughter when there is a rating difference of +63?

I have posted links several times. I'm on my phone right now, so I won't post any this time.

Look up the works of statisticians such as Regan, Guid, Bratzko (sp?) and others. Don't rely on the guesses of random chess.com posters.

Lol, I did not rely from random guessers from chess.com. I took reference from talkchess. which is the biggest computer chess programmers forum. e.g Programmers of  Stockfish, Houdini , Komodo.

Just find a 4000, maybe AlphaZero? and pair it with a 3000, maybe choose one from the TCEC Division 1?

How so?

There is still no 32 piece tablebase. Draw is just one guess. The other is White win. It should be a big news if it were proven either way, finally.

I dont think he's saying it has been proven. He's saying it's theoretical. So it's a theory, a guess. Other guesses include white can force a win (as you said), black can force a win (which is probably unlikely) and one side can force a draw while the other cannot (like if white always wins or draws but never loses, also probably unlikely).

if you are seriously wondering if chess is a win by white merely on the lack of 32 piece databases, then might as well wonder if black is a forced win as well. it could all be one giant case of prophylaxis after all!

its called a conjecture. something virtually everyone knows (chess is a draw) but we have no way to  formally prove it (yet at least)

People know chess is likely draw, only when they have considerable amount of knowledge, or experience.

If you ask  beginners, they may think  anything can happen.

If you ask any of top 2600-2800 players, almost all of them will say draw. 

Robert Houdert once said in TCEC chat that " Houdini 6 is much better than Houdini 5 but Houdini can win only when his opponent does mistake". (These programmers have extensive amount of chess experience, he tested for 20 million games between Houdini 5 and Houdini 6)

I dont think a win without at least one mistake from the losing player has ever happened. Even in stockfish vs Alphazero games there were mistakes, though hard to identify. So, from a logical point of view, perfect play=draw, because if one side loses then didnt play perfect.

So you have two sets of beginners, one who say anything can happen (which is probably a healthier approach) and another set of beginners who almost all say it's a draw. The first set of beginners say anything can happen because when they play, anything can, and does happen. The second set of beginners experience a very high percentage of draws, so they predictably will say chess is very likely a draw. Both sets of beginners say what they say based on experience. 

But what happens when the second set of beginners plays against opponents rated 800 points higher? From the comments I have seen here they never win, ever. They draw sometimes, but almost always lose.  Suddenly chess is not a draw because one side (the beginners) make huge mistakes. So it seems to me that speculation on perfect play is just speculation because no living person has any clue what perfect play is. How can a beginner say what perfect play is when not only can they not win, they cant even draw a machine that is likely to bet even better. 

Maybe 4000 in unachievable an maybe 3500 is as high as ratings will ever go but given the progress in the last 50 years it seem likely ratings will go above 4000 at some point, even if it's 200 years from now. At that point guesses on what perfect play are and guesses at what the outcome of chess are will be more credible.

nothing worse than training a class player content to draw. they get a pawn or even two pawns agaisnt someone 200 points above them and settle for a draw, or in simuls where they are winning! compound the problem if they only want to play "proper" chess with slow boring openings which further allows them not to fight as much for initiative.

Castrated players. 

A better way to teach beginner is " Statistical chess" rather than "brute search chess."

Stockfish and 99% other common engines are using Alpha Beta(AB engines) brute search to find a solution.

The main problem with those AB engines is that they dont know the probability.

Stockfish will say "0.00" when he see 1 out of 100  line is draw( while other 99 lines are losing). 

There is another line where there is 99 out of 100 line is winning but one line is draw, but stockfish will say  the same "0.00"

So stockfish will see moves as the same score and will  play either of those move randomly or the last searched line from his cache memory.

But in practical play, you will likely lose if you play 99% chance of losing line.

That always make crazy to human players of any level, except for  programmers.

That is why people will say " Dont follow computer lines, they are dangerous", that is practically true due to these reasons.

On the other hand, there are Monte Carlo Tree Search Engines(MCTS) that use probability in choosing moves,

In above two examples, although stockfish says , 0.00 for both, MCTS engines will say 1% for the first move and 99% for second move, and will always choose the second move, that will likely get more chance of success in practical.

The top MCTS engines are Alpha Zero, Komodo MCTS , Leela Zero, Scorpion 2.8.

I suggest beginners or any level of human(especially beginners to intermediate) to use MCTS engines evaluations, for practical chance. ( The only problem with MCTS engines is that they are much weaker than AB search engines though)

From these example, although chess is likely draw, a person who choose more probability of winning will likely win!!!

There is something worse than coaching a player who is content to draw.

I have seen scholastic coaches who *teach* "fireproof openings" to their students. They teach their students to fear their opponents and play it safe. The students are taught to be content with a draw, and that chess is boring. Such players usually drop out of chess entirely after a year or two.

[post to attempt to fix broken forum format]

[succeeded, it seems]

Yes the 3000 could attain 4000 ayesdeeef, since he could beat every single person in the world and easily obtain his elo over 4000