Upgrade to Chess.com Premium!

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


  • 3 days ago · Quote · #2721

    Nekhemevich

    yeah. What's the glicko?

  • 3 days ago · Quote · #2722

    Nekhemevich

    referring to potential player winning a gm, which is great, but is that where the story ends?

  • 3 days ago · Quote · #2723

    Ziryab

    RetiFan wrote:

    Of course, I'm talking about games when both players want to win.

    I also don't buy %0 percent chance, because I think I can get a win against a Boris Gelfand type blunder.

    The OP's op.

    The chance is better than zero. There's a 0.000000000000000000000000000000000000000000000000000000000000001% chance, approximately.

    I may be short one or two zeros. 

  • 3 days ago · Quote · #2724

    dpnorman

    @Stephenson2 it was a tournament hosted by one of the chess clubs I go to. There is only one section, and the player strength within the section usually ranges from a "floor master" (a national master who has lost it a bit as he's aged and now has trouble staying above his 2100 floor- I've only played him once but with any luck he will soon become my highest-rated victim haha) to a 1100 type player. There are a bunch of 1600-2000 players there. But yeah if you're playing in a major open tournament, that sort of upset would only be possible with someone playing about two sections up.

  • 3 days ago · Quote · #2725

    Elubas

    dpnorman: Because you would probably have to win 30 games in a row to get your rating up that high in that short a period of time from a non-provisional rating :)

    I actually think the chance of a 1300 beating a 2700 is higher than most people do; but I try to stay unbiased. When I see claims of people beating 2000s as 1400s, I will certainly investigate the most plausible explanations first.

  • 3 days ago · Quote · #2726

    Ziryab

    We established the number of zeros after the decimal a few pages back. Maybe page 40. Maybe 25.

  • 3 days ago · Quote · #2727

    dpnorman

    @Elubas Nope. I've just been playing a lot of higher-rated players and having results above my rating in my tournaments. If I had won 30 games in a row against higher-rated players, I would be much higher haha.

  • 2 days ago · Quote · #2728

    Linc95

    If anyone has heard of Wolfram|Alpha, it has a win percentage chance calculator - just input the ELO ratings you'd like to compare. It estimates the likelihood of a 1300 winning a match up vs. a 2700 to be 0.0316% (roughly 1/3164). Bear in mind it's a theoretical calculator only! In reality I would personally expect the odds to be even lower than that, I just thought some of your folks minght find it interesting.

    Link: http://goo.gl/NLrlCk

  • 2 days ago · Quote · #2729

    Nekhemevich

    As far as actually sitting across the board from a really great player, and it happens to be a 2700!!! I would be very grateful for the time and less concerned of winning as I am of learning.

  • 2 days ago · Quote · #2730

    Ziryab

    Linc95 wrote:

    If anyone has heard of Wolfram|Alpha, it has a win percentage chance calculator - just input the ELO ratings you'd like to compare. It estimates the likelihood of a 1300 winning a match up vs. a 2700 to be 0.0316% (roughly 1/3164). Bear in mind it's a theoretical calculator only! In reality I would personally expect the odds to be even lower than that, I just thought some of your folks minght find it interesting.

    Link: http://goo.gl/NLrlCk

    Analysis of actual results have demonstrated that the higher player underperforms relative to Elo predictions up to about 500 Elo difference, but then after a 500 point difference, the higher rated player performs significantly better than the predictions.

  • 2 days ago · Quote · #2731

    Nekhemevich

    Well... I'm happy I put my time to good use. Until next thread... however long that takes. Hopefully you can all get through standard deviation class, as far as me I'm going to play chess. :)

  • 33 hours ago · Quote · #2732

    BMeck

    Linc95 wrote:

    If anyone has heard of Wolfram|Alpha, it has a win percentage chance calculator - just input the ELO ratings you'd like to compare. It estimates the likelihood of a 1300 winning a match up vs. a 2700 to be 0.0316% (roughly 1/3164). Bear in mind it's a theoretical calculator only! In reality I would personally expect the odds to be even lower than that, I just thought some of your folks minght find it interesting.

    Link: http://goo.gl/NLrlCk

    I would love to see the data they use to calculate it because I dont believe it is that "high."

  • 33 hours ago · Quote · #2733

    SweetSurrender

    I, as a 776 player, once beat a 2700 chess master .... of course, we were playing horseshoes!

  • 33 hours ago · Quote · #2734

    Linc95

    Ziryab wrote:

    Analysis of actual results have demonstrated that the higher player underperforms relative to Elo predictions up to about 500 Elo difference, but then after a 500 point difference, the higher rated player performs significantly better than the predictions.

    That's an interesting point. I haven't come across any analyses on that before but I'll take your word for it. The calculator certainly wouldn't be taking into account differences in contextual performance like that^... I wonder how its predictions would change if it did? Very interesting :)

  • 33 hours ago · Quote · #2735

    Ziryab

    There both a citation to and discussion of a research article that presents the data a few pages back in this thread. It was maybe six months ago. There's also a link to the article.

  • 33 hours ago · Quote · #2736

    Linc95

    BMeck wrote:

    I would love to see the data they use to calculate it because I dont believe it is that "high."

    I definitely agree with you, the chance is probably even lower, however that calculator is based on the equations underpinning the ELO rating system not an actual analysis of data. It's mathematical and conceptual only - reality is definitely going to be different! If we analysed proper data I'm sure the chance would be even lower than the ~1:3164 it predicts but for the sake of simplicity and for anyone interested I thought that calculator was relevant (even through it's probably not 100% accurate). As we seem to agree, reality would probably give an even smaller win-likelihood to the poor 1300.

  • 30 hours ago · Quote · #2737

    Elubas

    "As we seem to agree, reality would probably give an even smaller win-likelihood to the poor 1300."

    Well, it seems just as speculative to insist the prediction is false as it is to insist that the prediction is true.

    It pretty much does just come down to the number of mistakes players make in the end. A 2700 doesn't make a lot of them; that's what got them there. A 1300 makes a lot of them; that's what got them there. The harder the opponent for the 1300, the more this gets punished. The task becomes harder and harder for the 1300, but I don't see why this can't be measured mathematically. So for a 1300 to beat a 1500, they would have the same difficulties as say a 1400, but even more so. I don't really get why this logic changes if we kept going further. It's not like the value of 100 points increases as time goes on. So the 2700 suddenly plays like a 3500 when playing a 1300? No, their skill is just obviously much higher than the 1300, which is why the prediction is as low as it is (1 in 3164 is extremely low).

    It seems like people want to inflate the strength of the 2700 just because the encounter is lopsided. If the difference in strength was really that high then the rating system would stretch out more. You'd have people who really do have a 1 in 3164 chance of losing to a 1300; then those people would be the 2700s, and maybe the Carlsen's would be at 4000. In any case 1 in 3164 is virtually unbeatable anyway.

  • 29 hours ago · Quote · #2738

    DjonniDerevnja

    Linc95 wrote:

    If anyone has heard of Wolfram|Alpha, it has a win percentage chance calculator - just input the ELO ratings you'd like to compare. It estimates the likelihood of a 1300 winning a match up vs. a 2700 to be 0.0316% (roughly 1/3164). Bear in mind it's a theoretical calculator only! In reality I would personally expect the odds to be even lower than that, I just thought some of your folks minght find it interesting.

    Link: http://goo.gl/NLrlCk

    I dont believe it.1300´s have good and bad days, same with the GM. 

    Magnus Carlsen blundered against Anand in a WCgame, and Anand blundered back, he didnt see it. The GM´s are not overhuman supergods. They too make mistakes,

    and a skyrocketing 1300 on the way up (maybe actual strenght closer to 1900) does play at his best some days, and that can be fantastic chess.

    One day a 1300 can play as a 2000, and at a very bad day a GM can play as a 1800. A Gm can also test a line he doesnt know well yet, and be punished on that.

    I had a game when I was at878N-Elo , fideunrated, when I with black outplayed a 1800fide, and mated him in 78 moves. I played fantastic, and he wasnt bad. His mistake was that he did not try to play for a draw when I opened absolutely smooth. That day I was better. I played like 2000. A GM could have done a similar mistake.

    Most players are theirselves worst enemy. If they are playing their best, they can beat almost anybody. If not, they can loose to almost anybody.


Back to Top

Post your reply: