< Earlier Kibitzing · PAGE 3 OF 3 ·
Later Kibitzing > |
Apr-16-07 |
| barbababa: <Plato><The ratings indicate performances over time, and to have a 50 point advantage at the GM level is much more significant than having a 50 point advantage at, say, the class B level.> Actually, 50 points ELO rating advantage should be as significant in any level. For example, if 2700 player plays a 100 game match against 2750 player, the result should be the same on average as if 1700 player would play a 100 game mach against 1750 player (43-57). |
|
Apr-16-07
|
| kellmano: What a cool dude this guy is. I'm gonna start supporting him: http://www.chessbase.com/newsdetail... |
|
Apr-16-07
|
| Plato: <barbababa> In my opinion differences in rating are more significant among GMs than amateurs. A 50 point difference at the GM level tells us more about their respective levels of understanding than a 50 point difference at the "patzer" level. This may be partly because GMs tend to play more frequently than amateurs, so the data for them is more reliable. I would contend, for example, that the difference between Kramnik (2772) and Navara (2720) is greater than the gap in strength between a typical 1772 and 1720 player. I know that this is just one stark example, but I'm of the opinion that it applies in most cases. |
|
Apr-16-07 |
| stanleys: <Plato:><At that level 50 points is quite a lot.> This is not so sure.Take the example of Sutovsky whose rating 2 or 3 years ago was 2697,but at the end of 2006 he was rated 2607.You could call him a super GM in 2004 and a modest one in 2006,but I think it's the same guy. In these championships there are many underestimated players between 2500-2600(or little higher).That's why the "stars" prefer to avoid playing in them - they know that the risks are too important,and the success in not guaranteed |
|
Apr-16-07
|
| Plato: I agree with some of what you say. There are definitely some underestimated players in tournaments like these. But I still think a 50 point difference at the 2600+ level is significant. If we judge by head-to-head scores and common tournament results, a 50 point rating gap among "the stars" tends to reveal a clear difference in their relative strength -- more so, I believe, than among low rated amateurs. Consider Anand and Svidler (50 points) or Ivanchuk and Malakhov (50 points) or Ponomariov and Sokolov (54 points). There are always some exceptions in cases when one player is very much overrated or another is very much underrated, but in general the ratings of GMs are more reliable than the ratings of amateurs because they tend to play more regularly (more data is available). The result is that rating points are somewhat more indicative of actual strength at the GM level than the amateur level. The ratio of games in which 1600 players defeat 1700 players is almost certainly greater than the ratio of games in which 2600 players defeat 2700 players. Of course Sutovsky is the same guy, but it's not obvious that he was playing as well at the end of 2006 as he was two years earlier when he was rated 2697. How can you be certain this is the case? In any event, Sutovsky's case is more of the exception than the norm... I am certain that if we examine rating fluxuations over a certain period, it will be evident that the ratings of amateur players tends to fluxuate by wider margins more frequently than the ratings of 2600+ GMs (at least when the number of games played is comparable). |
|
Apr-16-07 |
| stanleys: <Plato:>
Of course you're right about the amateurs,here I have to disagree with <barbababa> |
|
Apr-17-07 |
| esticles: An important fact that has not yet been mentioned in this discussion is that longer Swiss-system tournaments disadvantage players who start off well. This may seem paradoxical but it is true--players who play well in the early rounds always face the hardest opposition, and in longer tournaments the early leader (even if he is the highest rated) hardly ever ends up on top. This tournament's example is Volokitin, who led early on and finished with only 7.5 despite a performance rating of 2749 (25 points higher than Tkachiev's!) You will find similar "victims of the Swiss" in any tournament of extended length. So be careful what you wish for. Making a tournament of this size longer than ~9-11 rounds will actually INCREASE the likelihood that a lower-rated player will win. |
|
Apr-17-07
|
| Plato: <So be careful what you wish for. Making a tournament of this size longer than ~9-11 rounds will actually INCREASE the likelihood that a lower-rated player will win.> Not at all. The longer the tournament lasts the better the chances that the strongest player will win it (the highest rated player and the strongest player are not always the same, of course). Statistically it's the only thing that makes sense; as the number of games increases the results tend to become more accurate reflections of strength. Also, the more rounds there are the more common opponents that the top finishers will have played, so longer tournaments have the benefit of counteracting the problem you mentioned of one player facing tougher opposition. |
|
Apr-17-07
|
| JointheArmy: Just a note on <barbababa>'s statement, you have to take into account that FIDE changed the K-factor for players over 2400 to a ridiculously low K=10. Amateurs have a higher K-factor which means their ratings fluxuate more and it takes a shorter time for players to gain or lose points. So it would take longer for let's say Bacrot to gain 30 points than it would for amateurs. I heard FIDE changed the K-factor again in 2004 which explains the slight inflation with the top 100 being ~2620 instead of 2600 flat. |
|
Apr-17-07
|
| Plato: As a side note, the 13-round format *was* used from 2001-2005, and I hope they go back to it after what transpired here. I just feel it's ashame for the title of European Champion to come down to a seven-player rapid K.O. Tkachiev was definitely the best rapid player at the tournament, but was he the best classical player of the top finishers? In my opinion he was not. Having a couple more rounds would have greatly increased the likelihood that the best player (of "real" chess, that is) would have been crowned European Champion. |
|
Apr-17-07
|
| JointheArmy: <esticles> Yes and there was a report by ICC backing up your claim. Also Akopian's latest Swiss victory backs up the claim. I hate to bring up ICC again, but ICC insists the lower ratings in the 1-minute and 5 minute categories are due to the auto-pairing format heavily based off Swiss events. For example you play an opponent that has won the same number of games in a row you have won or something similar. I'm not sure I agree with this, but its very hard to argue against the numbers and the world's largest chess site. :-) |
|
Apr-17-07 |
| esticles: <The longer the tournament lasts the better the chances that the strongest player will win it.> Not quite. The longer the tournament lasts, the better the chances that the highest rated player will have the highest performance rating. But the longer the tournament lasts, the better the chances that the person with the highest performance rating will NOT be the winner of the tournament. Of course, this trend is only true up to a point, because there are a limited amount of players. As you pointed out, with very long tournaments the top players will soon all be playing the same opponents and the tournament will be closer to a round-robin (a much better gauge of ability). But at 13 rounds (with this tournament size), this phenomenon is not enough to reverse the increase in difficulty posed to players who do well. A 19-round tournament might do the trick, but this is of course impractical. |
|
Apr-17-07
|
| Plato: If one compares longer Swiss events with shorter swiss events (in which a similar number of players participated), I am convinced that higher-rated players win the former more frequently than the latter. It's true that in Swiss Systems the players who start out strong tend to meet the toughest competition overall (and this is only right), particularly if the tournament is too short. But that's just another reason why longer tournaments tend to be more accurate reflections of strength. |
|
Apr-17-07
|
| Plato: <But at 13 rounds (with this tournament size), this phenomenon is not enough to reverse the increase in difficulty posed to players who do well.> Players who do well are generally going to face increasing difficulty. That's the way it should be. Thirteen rounds would have made a significant difference... Examine what happened. Let's assume for the sake of argument that Tkachiev was *not* the best classical player at the tournament, even though I think most would agree (even before the playoffs) that he was the best rapid player. In the main tournament Tkachiev had five full points, four were OTB wins and one was because his opponent forfeited. Each of the players he defeated was rated in the 2400s-2500s. This was at least a slightly less arduous road than, say, Jakovenko's or Pavasovic'. With two more rounds Tkachiev would have had to play to more tough opponents, guys like Sutovsky or Jakovenko or Cheparinov or Sakaev. It's clear that with two more rounds the stronger players would have more time to establish themselves. Of course I would agree that Tkachiev was one of the strongest classical players in the tournament, but I don't believe he was *the* strongest. Two more rounds would have reduced the likelihood of multiple-ties and increased the likelihood that the winner would be the strongest classical chess player. <<The longer the tournament lasts the better the chances that the strongest player will win it.>Not quite.>
I don't really think this is debatable. It is a matter of sheer statistics. It doesn't make any sense to contend that the strongest player would be more likely to win an 11 round Swiss than a 13 round Swiss but less likely to win 11 rounds than 19 rounds. As the number of games increases that data increases and as the data increases the results become, statistically, more accurate reflections of relative strength. |
|
Apr-17-07 |
| stanleys: This guy is a real showman:http://chesspro.ru/_events/2007/eur... |
|
Apr-17-07
|
| JointheArmy: <stanleys> Tkachiev is such a badass. Seriously does anyone look cooler in chess than that guy? |
|
Apr-17-07
|
| Plato: He and his brother Evgeny are the "blitz brothers" who organized the "World Chess Beauty Contest": http://www.chessbase.com/newsdetail... |
|
Apr-17-07 |
| stanleys: He is entirely what the French people call "bon vivant" and it's nice to see.
Gives a good image to chess,I think |
|
Apr-17-07
|
| Knight13: *Holds up a cup of Coca Cola* To the Champ of Europe! |
|
Apr-18-07 |
| esticles: <With two more rounds Tkachiev would have had to play to more tough opponents, guys like Sutovsky or Jakovenko or Cheparinov or Sakaev. It's clear that with two more rounds the stronger players would have more time to establish themselves.> But Sutovsky, Jakovenko, Cheparinov, and Sakaev will also have to play more tough opponents. The people who do not are the people half a point behind. While it is very difficult for any of the leaders to get more than a half-point, probability says that at least a few of the 2550s in the 7.5 slot will beat another 2550 and vault themselves into joint first. The same thing will happen with the 8.0s in round 13, and so forth. <It is a matter of sheer statistics. It doesn't make any sense to contend that the strongest player would be more likely to win an 11 round Swiss than a 13 round Swiss but less likely to win 11 rounds than 19 rounds. As the number of games increases that data increases and as the data increases the results become, statistically, more accurate reflections of relative strength.> Again, you are confusing strength of play (think performance rating) with number of points. The two do not go hand in hand in Swiss events. To shed some light on this difference, say a near-infinite number of chess players played in a near-infinitely long Swiss tournament with no rules about playing the same person twice. After the initial place-sorting between stronger and weaker players, every single player would play every single game against someone with the exact same demonstrated strength. Therefore EVERYONE'S expected score is 50%. Of course the strong will have slightly higher scores (1.001 trillion vs trillion, for example) than the weak players, because of the initial place-sorting of the earlier rounds. But while the stronger players will continue to prove better in terms of ELO, they will not gain any tournament points on the weak players after the first few billion rounds. In an actual tournament, the strongest players are playing each other by round 7 or so (depending on the size). As in any tournament, the strongest DOES have the highest probability of winning. (I never disagreed with this.) But in a deep tournament such as the European Championships, this probability is no higher with 13 rounds than 11. The guys who are "almost as strong" will continually be slightly behind or equal to the leaders in terms of tournament score, and some will always be winning against slightly weaker players to jump into the lead spots. No matter how long the tournament, it will always come down to the last round. This continues until enough games have been played that the "best" stop playing the other "best" because they have already played them all. At this point, adding rounds begins to have a purpose again. Yes, it was dumb of me to arbitrarily say this happens at round 19--I have no idea when it starts (probably later than that). But the fact is adding rounds does have different effects on the tournament depending on the amount of participants and amounts of rounds already in place. So I maintain that IF the strength of the field is sufficiently deep, a 13-rounder will not increase the probability that the best person will win. I do concede that it will not decrease this probability, as I falsely stated earlier. Also, you are right if there is one player good enough to consistently beat (not just draw) the other top players, 13 rounds will give him more time to show his stuff than an 11-rounder. In a tournament with so many superstars, though, the differences in the top players makes the advantage of a 13-rounder over an 11-rounder pretty much negligible. |
|
Apr-18-07
|
| Plato: <esticles> You make some good points. I still have certain disagreements regarding a couple of your comments, but rather than getting into a protracted mathematical debate on the page of the newly crowned European Champion, I think it's best we leave it here and agree to disagree on a few minor points. |
|
Apr-18-07 |
| dehanne: And this is how Tkachiev became European champion, doing the "Danailov" : http://chesspro.ru/_images/material... ;-) |
|
Sep-12-07
|
| ahmadov: Tkachiev is currently in Baku to play 6 rapid games against Vugar Gashimov at the end of the women's tournament... |
|
Sep-13-07
|
| ahmadov: I have to correct my mistake. Tkachiev and Gashimov played 6 blitz games, which ended 3:3... It was amazing to watch them playing so fast... |
|
Oct-27-07
|
| Karpova: He is the Ajaccio 2007 blitz champ!
http://www.echecs.asso.fr/ (you have to search around. the exact link doesn't work) 1. Tkachiev 25.5/32
2. Fressinet 24.5/32
3. Karpov 24.0/32
There were other strong players like Bacrot (21.5), Bareev (21), van Wely (20.5) and Milov (19) among others. |
|
|
< Earlier Kibitzing · PAGE 3 OF 3 ·
Later Kibitzing > |