Because of all the chaos and controversy sparked by the recent changes in the World Championship Candidates Cycle, I thought I would weigh in with my own ideas about how it should be run (This being the internet after all).
Bearing in mind that the sole purpose for the World Championship is to find the best player at a given moment in history. These moments have lasted less than one year (when the previous Champ won the rematch) up to 11 years (Between the Lasker Steinitz match and the Lasker-Tarrasch match). 3 years has been standard (Actually, not counting the FIDE matches and tournaments the average is more like 2.9 years. Interestingly, if you do count the various FIDE formats, then the Anand-Kramnik match was the 50th ever world championship).
But the systems introduced have been rather lax at producing the best. We had Lasker sticking around when Capablanca was dominating everything in sight. We had Kramnik champ when Kasparov still had 5 years of being the absolute top-dog left in him. In fact, looking at the only historical chess rating system worth looking at (chessmetrics) shows that the reigning world champ has only had the top performace rating at the end of the year 43 times out of 118 years (1886-2004). (The years they won the title are counted for them, but not the years they lost it).
Steinitz 1/7
Lasker 11/26
Capablanca 0/5
Alekhine 7/16
Euwe 1/2
Botvinnik 2/10
Smyslov 1/1
Tal 1/1
Petrosian 0/6
Spassky 0/3
Fischer 1/3
Karpov 6 out of 8
Kasparov 11 out of 13
Kramnik 1 out of 4
Kasparov and Karpov were by far the most `rightful`Champs. Petrosian should have held the title in`62 and `69 but not in between.
My system is inspired in part by this, and the accompanying article written by Jeff Sonas (For whom it was merely a thought experiment) and in part by a letter written to chessbase, I`m afraid I can`t remember who by, but I think they were proposing a new candidates system.
So, to my idea. From the preamble I imagine a number of you have already figured it out. The world champion of any given year should be the person who has the highest performance rating over the entire year, more or less. I`ve come up with a few caveats that I think ensure fairness.
The main advantage of course is that it wouldn`t allow a match between, let`s say Anand and Kramnik when Topalov, Carlsen and Ivanchuk are all playing the best chess of their lives and Vishy and Kramnik are turning in results that could be described as, at best, mediocre.
So what do you think. Are you happy with the current system, have ideas of your own, are you ready to march to Kalmykia in protest.
Very good article written
Thanks but I was more curious about what people thought.
I liked the system as it was in Fischer's day. One of the biggest problems is that the best tournament player often isnt the best match player. But I think head to head matches is a better way to determine the best player than a tournament. Petrosian was a great match player but not so good in tournaments due to his style. Geller was a great tournament player but did poorly in matches. To win tournaments you have to take more risks than to win a match thus the two formats are completely different. If you win a couple of games and draw the rest in a tournament you arent likely to win the tourney but do that in a match and you win . Petrosian deserved to be world champ from 63-69 because he first defeated Botvinnik and then Spassky in their first match. He may not have been the highest rated player and his style of play meant he didnt do so well in tournies but he rarely ever lost ! Spassky won their second match and also deserved the title.
OK, I applied the system to the 2008 results for the top 5 players (I'll probably do it for Jakovenko, Radjabov, Aronian and Movsesian later)
I didn't drop the worst performance, but I didn't count any game against players rated below 2500.
Carlsen - 2953 (3 seperate 2900+ performances put him over the top)
Morozevich - 2920 (His second place finish (Of the 5 I calculated) was against an average opponent of 2682.4, the lowest of the 5.)
Ivanchuk - 2888 (With 138 games, he was by far the most active player.)
Topalov - 2878
Anand - 2797 (With only 42 games, he was the least active player, but his average opponent was an incredible 2733.9)
OK, revised performance rating from the 8 players I thought would have the highest performance ratings (Well, I obviously wasn't going to do this for them all). As you can see from below, Movsesian, Carlsen and Ivanchuk were definitely 1st 2nd and 3rd. Morozevich was probably 4th, but I can't be sure because I didn't do anyone. I know Radjabov and Jakovenko had higher TPRs than Topalov and the rest, but due to an error with the first batch (I counted January '08 and didn't count the Jan '09 games, which took place in 2008 and 2009 respectively) I didn't redo it for Radjabov and Jakovenko, but neither were within even a 100 points of Ivanchuk.
1. Movsesian 3000 (HIs average oppoennts were only 2625, by far the lowest)
2. Carlsen 2975
3. Ivanchuk 2940 (With 128 games, he was the most active
Morozevich 2908
Topalov 2847
Anand 2812 (His 2749 average opponent, thanks in large part ot the macth with Kramnik, was the highest, he was also the least active with only 50 games)
Aronian 2806
Leko 2762
People who probably deserve to be checked are MAxime Vachier-Lagrave, Wang Yue, Boris Gelfand (For his Olympiad perf.) and Grischuk.
OK, so I decided that the FIDE calculator just won't work for this. I haven't been able to find the formula but it very strongly rewards actvity, but it goes up to a maximum of 99 games. This skews the numbers. So I'm re-entering the data and adding in Gashimov (Chosen for his Grand Prix win and rating increase).
The new calculator was chosen for impartiality ( it adds no bonus) and can be found at:
http://georgejohn.bcentralhost.com/TCA/perfrate.html
Name Score/ Average PR
#of Games Opponent
Ratings
These new results seem more reasonable to me, especially the lack of 2800+ TPRs and Topalov in 1st, who had an incredible string of victories, and Movsesian closer to the bottom, who was only chosen for consideration because of his rapid rating rise. Note that no bonus has been applied, as I'm rethinking the 0.25 points per game, though if they had been added Carlsen would have won. Here are the changes form old to new:
Movsesian 1-7 (-6)
Carlsen 2-2 (0)
Ivanchuk 3-5 (-2)
Morozevich 4-3 (+1)
Topalov 5-1 (+4)
Anand 6-4 (+2)
Aronian 7-6 (+1)
Leko 8-8 (0)
Gashimov N/A
The average change was 2 spots. Jakovenko and Radjabov will be added later, finishing the current top 10, perhaps Dominguez or or Vachier-Lagrave who both experienced rapid rating growth indicating high PRs.
I'll also be updating the current ratings to take into account a drop of the lowest Performing Tournament (I've decided to avoid the troubles involved in things like the Bundesliga (Notorious for elite players underperforming) by simply ignoring them. I expect the rankings to remain largely the same with the most benefit probably going to Anand and Topalov (Who has relatively few games, and scored 6/13 (a large denominator) in Corus.
New Rankings with Radjabov, Bacrot, Kramnik (Can't believe I forgot him. OK, maybe I can, but 2008 was not his year) and Jakovenko added:
There's the top 10, plus 2 (Gashimov and Aronian by the Jan '09 list.) All tournaments included of course (Dropping lowest PR is next) but all games against players rated below 2500 excluded.
Here are the rankings with each players lowest PR dropped. If this particular mthod of calculation i sused, then there should definitely be some sort of activity bonus, because of course, lower number of games means each PR has a greater affect on the annual PR, and will therefore get a greater bounce from this.
Some additions. Dominguez, who I thought had a much better year than he did (He was harmed by the fact that he had more games than anyone else against players below 2500), Grischuk (To fill out the current top 10, sadly his Linares Performance isn't counted) and Yue (Who was =1 in one of the Grand Prix events etc etc).
Worst Performance Adjusted Rankings
I like the head to head format for determining a champion. Your system is interesting, but I wonder if it couldn't lead to manipulation. People anxious to claim the prize would play as many games against ~2500 players as possible, as well as other manipulations.
well I chose to ignore games againt below 2500 players just for that reason. And even the best players in the world can't win every game against someone rated 2500, especially if the other GM is aiming for a draw as they ofen would against tough opponents. And opponents average ratings is very important in determining PR.
Also, this system could simply replace the various candidates systems hy taking the top player and having him play the previous champ, or the top 3 and having them player a couple of matches etc etc...
I just think that the Performance rating system should be taken into account in order to determine the champ because the title should go to the strongest player. And it's no longer clear who that is.
3 new ones, Shirov, Bacrot and Nakamura, Nakamura's adjusted rating will surprise you.
OK, so Nakamura's example of having a low number of games and his lowest performance coming from his longest event (8 rounds) allowed him to have a much larger boost. So there definitely should be an activity bonus, but I'm not sure how much. I think instead of just adding a certain number of points per game, I'm going to adjust FIDE's performance rating calculator, which is inherently favourable towards active players. But until then, Topalov is the champ.
So, to anyone who might actual read it (And to my future self for posterity's sake) I've been rethinking my approach to this. First of all, the ban on counting games with opponents below 2500, I think that has to go. I did some math, and if your opponents average rating is below 2500, the highest PR you can get (With a perfect score that is) is 2700. Not even high enough to make the top 20. With that in mind, the first change I'm making is to abolish this limit. Weirdly, despite playing 128 games, Ivanchuk only had one more game included.
Secondly, I think the Nakamura example showed us that we can't just dump a whole event, it's unfair to thos ewhose worst event was a single game (As in the Bundesliga or a tiebreak), so instead, everyone get's to drop their worst 10% consecutive games! ISn't that fun! Basically, if you played 100 games, we diregard 10 in a row, the worst 10 in a row. Topalov, for example, has his last 6 games at Linares disregarded. This should balance it out, but I'm still gonna keep the unadjusted chart around, in case it ends up being weird. I like this because it subtly rewards activity, and doesn't punish one horrible event.
The old lists are in the post right above this
Name Average PR
Opponent
p.s. I'm tired, I'll finish it later.
p.p.s. All Hail Magnus Carlsen, the newest World Champion.
p.p.p.s. Nope, not kidding.
Join Chess.com for free to add your comment! Already a member? Then login now to comment.