# Is there any chance that a 1300 rated player can beat a 2700 rated player?

• 3 years ago · Quote · #941
rooperi wrote:

Find all games where 2700+'s lost. Including overlooked mates. queen blunders etc.

Could  a 1300 realistically have put together any of the winning games?

If so, show me.

With quadrillions of "tries," possibly. Otherwise, no.

• 3 years ago · Quote · #942
Elubas wrote:

All of the above are great reasons why the chance would be extremely low -- I don't disagree these put the odds laughably against the 1300; I'm just saying that this doesn't mean zero chance. Let's say there is a 1 in 10000 chance the 1300 reaches the Gelfand position or something of the same level of advantage for white. And even here, we'll say that it's extremely unlikely that the 1300 will even win that. However, if he has even a 1 in 5000 chance of winning that position, something, then eventually, the 1 in 5 million shot may occur with enough tries.

Of course, this comes with the assumption that both components, the scenarios I have marked 1 in 10000 and 1 in 5000, are possible -- that's what I believe, for reasons given in earlier posts. If that is the case, then they simply have to both occur at the same time for the upset to happen, represented above.

What changed? Now you agree it won't happen?

• 3 years ago · Quote · #943

I'm not sure on the number -- could be 1 in 10000, could be 1 in 1000000000000000.

• 3 years ago · Quote · #944

Scottrf lol :-)

• 3 years ago · Quote · #945

Elubas probably something in between these two...

• 3 years ago · Quote · #946
Bluebird1964 wrote:

I had not seen that game before, what was Karpov thinking about? Guess you would have to ask him.

Apparently it was a 'blindfold' game.

• 3 years ago · Quote · #948

I really doubt this would happen often, but maybe as a miracle?

• 3 years ago · Quote · #949

So many posts, so little information.    The answer is yes.   There is about 0.032% chance of a 1300 player beating a 2700 player.

That's roughly a one-in-3000 chance, which is a whole lot better than any state lottery (thousands of times better!) and even better than your chances of getting a four-of-a-kind in 5-card-stud poker (which is about 1/4,000), much less a royal flush (which is much worse, at 1/650,000).

For reference and information on how to make these calculations yourself, look up "Elo Rating System" on wikipedia.

• 3 years ago · Quote · #950
GrandePatzer wrote:

So many posts, so little information.    The answer is yes.   There is about 0.032% chance of a 1300 player beating a 2700 player.

That's roughly a one-in-3000 chance, which is a whole lot better than any state lottery (thousands of times better!) and even better than your chances of getting a four-of-a-kind in 5-card-stud poker (which is about 1/4,000), much less a royal flush (which is much worse, at 1/650,000).

For reference and information on how to make these calculations yourself, look up "Elo Rating System" on wikipedia.

There's a whole bunch of better information than "look in wikipedia"

• 3 years ago · Quote · #951
GrandePatzer wrote:

So many posts, so little information.    The answer is yes.   There is about 0.032% chance of a 1300 player beating a 2700 player.

That's roughly a one-in-3000 chance, which is a whole lot better than any state lottery (thousands of times better!) and even better than your chances of getting a four-of-a-kind in 5-card-stud poker (which is about 1/4,000), much less a royal flush (which is much worse, at 1/650,000).

For reference and information on how to make these calculations yourself, look up "Elo Rating System" on wikipedia.

Once the rating difference grows much beyond 400 points, the equations cease to be accurate predictors of observed outcomes.

• 3 years ago · Quote · #952

According to FIDE, if the difference between two player's ratings is 677, the stronger player has 99% chance of winning

If difference is 800, the chance of winning is, for all intents and purposes, 100%.

Here's a link to their handbook with the relevant table:

• 3 years ago · Quote · #953

Simuls aren't reated for a reason, FEDTEL ;)

• 3 years ago · Quote · #954

Here's what I don't like about the rating system:

If Magnus has the highest rating of all time, does that mean he's the greatest of all time, or the greatest among his generation?  I think the rating itself is more a relative term than an absolute figure, so comparing players at their peaks based on rating isn't worthwhile.  If you get right down to it, the rating system is flawed when you put it that way.  I suspect the number will keep inflating, and the potential expansion of GM titles is indefinite.

Super duper bad mamma jamma grand master!

• 3 years ago · Quote · #955

Here's what I don't like about the rating system:

If Magnus has the highest rating of all time, does that mean he's the greatest of all time, or the greatest among his generation?

No. Ratings are relative only to the rating pool in which they are established.

I think the rating itself is more a relative term than an absolute figure, so comparing players at their peaks based on rating isn't worthwhile.

However, comparing players to their peers based on relative difference in strength does translate moderately well across eras. That's why Fischer's rating dominance of over 100 points above Spassky is really still more significant than Kasparov or Carlsen's achievements.

If you get right down to it, the rating system is flawed when you put it that way.  I suspect the number will keep inflating, and the potential expansion of GM titles is indefinite.

Super duper bad mamma jamma grand master!

The rating system isn't flawed, it simply isn't meant to give one an historical ranking of players across eras. That's like saying a refridgerator if flawed if it isn't also a TV. While it might be a great idea, it isn't a flaw that it lacks that feature. Rather, it simply means it wasn't designed that way.

• 3 years ago · Quote · #956

Just because Carlsen's rating is relative to his player pool, doesn't mean his rating is necessarily inflated compared to other generations. It could mean that, but not necessarily.

• 3 years ago · Quote · #957
Elubas wrote:

Just because Carlsen's rating is relative to his player pool, doesn't mean his rating is necessarily inflated compared to other generations. It could mean that, but not necessarily.

If rating inflation is real, it definitionally means precisely that, since rating is a reflection of performance, and if his rating is not inflated with  regard to a prior era, but the rest of the pool is experiencing rating inflation, then his rating would be higher as his performance would be even better than it has been since he should be scoring better against the opposition than he is.

Or perhaps more simply: ratings are calculated relative to one's opponents. If rating inflation is real, then Carlsen's rating has been calculated based on inflated values. This means his rating must be inflated as the inputs to the formula are infalted.

It should be noted that there isn't universal agreement that rating inflation is real, or that if it is real, how big an impact it actually has.

• 3 years ago · Quote · #958

Maybe I am misunderstanding your post, but your first sentence is like saying "If A is real, then A must be real." Yes, but I'm not presupposing that A (inflation) is real. I am suggesting the possibility that his rating might be higher not because of inflation, but because of an increase in quality.

Of course, something that would prove rating inflation would be proving that the quality of Carlsen's moves are the same as of a previous generation, even with a higher rating. Some studies have shown that the intrinsic quality of the moves, not just rating, has gone up in general.

• 3 years ago · Quote · #959

Agreed. I think that the middle ground is actually the real story. There is rating inflation, but I think the effect and impact is over-stated. The real big differentiator isn't rating inflation, but the ubiquitousness of computer-assisted training, the quality of the engines and the databases, and the fact that there is a strong pool of still improving players at the top.

But I think the evidence for rating inflation is there as well, and I don't think it is easily ignored either.

• 3 years ago · Quote · #960
FEDTEL wrote:
Kingpatzer wrote:

Simuls aren't reated for a reason, FEDTEL ;)

why? do you mean that masters don't play seriously? I know that they may allow a draw sometimes just to finish the games and go back home, but a loss? I think that's too much for them

Lets say GM is 2700, playing a simul against 60 people who average at 1200.  Then lets say GM spends 15 seconds per move, making each turn 15 minutes long for the kids.  This is a huge time penalty for GM.  I once heard that the average game is considered 40 moves.    If all of the games go for 40 turns, this means the games last 10 hours, making it 10 minutes for the GM.  Whether the game draws or someone wins, the GM just cut his time down by 15 seconds per turn.

As you can see, with this simple hypothetical, this encourages GM to knock out as many of the kids as possible.  The weaker kids go first, either by a quick defeat or GM doesn't want to fight and offers a draw.  Then he gets some more time to deal with the more challenging players, but since he rushed the games, he might find himself to be up a certain creek without a paddle.  This is what leads to his losess.

I think this situation says that simul games shouldn't be rated.  Everytime GM draws against one of the kids, he loses hundreds of points.  GM would never do simuls, then, since the rating loss wouldn't be worth the money.  Of course, GM could just be paid for their 10 hours of work, which would cost a lot more money and therefore no one will fund the simuls.  Who wants to do 10 hours straight chess, either?