Suggestion: Eliminate K-Factor in ratings

Jump to forum:
 
29th November 2008, 05:11am
#1
by Zenchess
Omaha United States
Member Since: Nov 2007
Member Points: 391

Hi.  I've been using chess.com for over a year.

I haven't used chess.com much or took the games very seriously, but I did get my rating up to 1800 or so which is at least somewhere near my uscf rating. 

I had my rating up to 1800 by playing many games.  I then decided to stop playing chess on chess.com until I was going to take the games seriously.  I had one game going at the time so I went on "vacation". 

Recently, I decided to start playing chess on chess.com again.  Instead of using it as a source of entertainment, I'm going to use it as a serious training tool to improve my over-the-board tournament career. 

So I started continuing that game I had going and I'm now losing pretty badly.  I may be able to pull out a surprise draw or even win if I calculate almost every possible variation in my head, but most likely as I see it now I will probably lose, and very quickly. 

I have not played on chess.com for a looong time.  So my "K-factor" or whatever it is that increases the longer you are away from the site is, had grown very large.  So it's like when I came back, I was PROVISIONAL again. 

The funny thing is, I wouldn't be complaining about this if my adjourned game was against a bad opponent or I was winning, or if I hadn't played lenny.  In that scenario, I'm assuming I would be rated much higher now (like over 2000). 

So in reality, I joined a game with bongcloud because he announced that he was "back".  So I was really psyched about this because bongcloud was like a hero to me. 

I decided to take bongcloud's revolutionary chess ideas one step further: My king would not only get to the other side of the board, but it would "drag" his queen with it.  So in other words, the king would make one step away from the queen, then the next move the queen would follow along the same path.  It's a very simple 'snake' like movement system such as you would find in snake or tron-like games.

So, bongcloud makes his normal king journeying moves.  I, very happy at this time, begin my king's journey, and open a dialog with the legendary bongcloud.  I think I may have been a little too 'far out' for him because he didn't say much.

Somehow, his queen found its way to the third rank, and mated me on g6 out of nowhere.  My king was self-mated by his own queen.  It's almost like I was playing some kind of twisted self-mate problem where the most unlikely and ridiculous mate occurs in the shortest time possible.  Like basically what would happen if a master were playing his 5 year old son and for fun brought his king out to teach him about attacking it, and the kid played along but didn't know his queen could move more than 1 square at a time, and decided to bring both his troops up to the battle and cried in shock as his dad delivered checkmate and won the game. 

So, that made me laugh actually, until I looked at my rating. 

1528.  I went from around 1840, to 1528 in one game that was under 11 moves. 

Now I am going to lose another game, which is going to lower my K-factor even more, and I'll be more 'stuck' at 1500 having to fight my way back to my regular rating, which was under-rated anyways. 

So, it's also possible something else happened: When you lose to a player over 1000 points weaker than you, you lose 300 points of rating in one game.  If this is true (I'm assuming it's not at this point) imagine a grandmaster playing a charity tournament against a 1000 rated player, and because he was not paying attention, or didn't care, lost the game.  And then his rating skyrockets downward from 2800 (super gm) to 2500 (weak gm) level.  This would destroy his career. 

Er, sorry about that rant, just wanted to explain what happened. 

My suggestion is:  a) K Factor Scenario: Eliminate K-Factor.  Adopt the exact scoring system of the uscf and normalize the rating pools to accord to it.  It would be highly accurate in regular live chess time controls, somewhat inaccurate in correspondence time controls due to players who put in more time than others.

  The most you could win or lose in a single game would be 32 points after you have played your first provisional 20 games, just like uscf.  Therefore, after you are not provisional, you are never punished for not playing for a couple years.  Dust off your rust in the chess club, or online, and return to your chess games (which I have done many times in my 11 years of playing competetive chess) and you will keep the rating that you are probably stronger than anyways.  The 2 major benefits I see to this is: People don't have to have a heart attack about losing a game, because it's just 32 max points (and usually much lower).  Second, the ratings will roughly accord with USCF ratings. Since a revival in chess is possible in America (where it has boomed before!(and could be accomplished with the help of chess.com especially!)), chess.com would have a lot to gain from rough accordance to USCF rating scales. 

How is such a boom possible?  By chess.com endorsing Hikaru Nakamura, chess genius and american PRODIGY unseen in america since the days of Fischer.  According to Jeremy Silman, Nakamura's play is hard to understand.  And he didn't mean it in a negative way, he meant that he could understand Kasparov better than Nakamura.  This shows how amazing his talent is.  What is even more amazing is that Nakamura is going to college, plays poker, invests in the stock market, etc. 

Nakamura could bring major chess coverage in american news again if he simply won some serious events, which he is very capable of doing.  This is why he needs sponsorship - he needs to get in these big events, because he's had trouble in the past.  The tournament circuit on the very high level is very biased, political, and character-based.  Not all players get the same opportunity to play in major events where they can gain experience and money.

When Fischer played the Soviets he became national news, and went on talk shows.  There was an EXPLOSION in chess tournaments nation-wide.  People were just showing up.  It hasn't been anywhere close to that in years. 

If Nakamura became strong enough to win top dog status, with a little help he could get some media coverage as a new American Genius the likes of Morphy, Fischer, and bring the numbers back to chess.  It takes an entire system of balances to operate for chess to succeed in the united states.  It takes celebrities like Fischer, and many other things.  Right now the united states has a surpising number of chess programs in schools, but most drop out after school.  If chess became a national sport with Nakamura as the celebrity chess.com would have an interesting advantage in the chess marketplace as it had contracts and or marketing deals with a prized celebrity. 

In this regard of growing chess in America, a celebrity like Nakamura is essential, but equally as important and necessary is the infrastructure to allow the chess community to grow.  Since chess.com already owns key domain-name real-estate (what could be more valuable?), it stands in an awesome position able to rally the chess community into an organized whole.  Since it is also a social networking site, I would assume it would not be a huge leap of imagination to wonder if chess.com could open up forums for every chess community in the united states, for instance, an "Omaha Chess" community forum hosted at chess.com.  It's very easy to spread the word for that to a friend: You just say "chess dot com" and say "click on local".  The site then immediately puts you into the Omaha chess.com community forum (because it has your IP address of course, which it can narrow down to city) and you can begin communicating with people who play chess in your city that you never knew lived nearby. 

Thus, the tournament scene expands, tournaments begin offering bigger cash prizes, thus the tournament scene continues expanding, thus thus thus thus * thus the infinity thus, and then the chess expantion happens at an exponential UPWARD curve instead of an exponential DOWNWARD curve which I've seen happen to so many games I've played over the years that it's going to drive me crazy if it happens to chess.

Chess.com = chess = tournament chess = chess teachers = chess authors = chess grandmasters = advertisers = event organizers = chess teaches logic = society improves = its the ciirrrrccclleeee  the icirrrccleee ooff lliffffeee

b) I suggest we adopt the uscf rating system (or fide, that's better for multi-national).  It seems weird to me when there's a 64 or more point shift after one game. 

Thank you for taking my suggestions under consideration!

29th November 2008, 05:56am
#2
by pcfilho
Rio de Janeiro, RJ Brazil
Member Since: Jul 2007
Member Points: 229

I'll read it.

29th November 2008, 06:04am
#3
by Spiffe
Orlando, FL United States
Member Since: Sep 2008
Member Points: 954

Wow, that's a lot of post.  I actually read most of it, though I skipped some of your 7-paragraph non sequitur about Hikaru Nakamura.  Might've been better to stick to the point there.

As far as your arguments about the rating system, I disagree.  You took a long break, and when you came back, you immediately started losing and your rating went down significantly.  That sounds to me like a system working perfectly... perhaps your strength is not as high as you thought?  If it is, it will recover just as quickly, no worries.

29th November 2008, 06:29am
#4
by Zenchess
Omaha United States
Member Since: Nov 2007
Member Points: 391

I didn't 'start losing'.  I lost a single game to lenny bongcloud.

29th November 2008, 06:31am
#5
by Zenchess
Omaha United States
Member Since: Nov 2007
Member Points: 391

And as far as my strength not being as high as I thought, no that doesn't apply.  If anything I've gotten far stronger since I stopped playing on chess.com, as evidenced by actual over the board uscf-rated events in which I've gained tons of points and am now rated 1980 once my games catch up and will soon easily be over 2000 when some future games pan out.

Also all my live chess ratings on chess.com are pretty high as well.  So while I appreciate your suggestion, I don't think it fits the facts that I am playing strong as ever yet my single game loss did not mean I dropped 300 points in actual strength.  Anyone can lose a game. =)

29th November 2008, 07:38am
#6
by erik
Mountain View, CA United States
Member Since: May 2007
Member Points: 10092

with a high RD your rating can go up as quickly as it an go down :) just get out there and win a few! :)

seriously though, i appreciate the thoughts. i don't think we'll change it now, but good to hear an opinion on the topic.

29th November 2008, 07:56am
#7
by staggerlee
Clermont-Ferrand France
Member Since: Mar 2008
Member Points: 790

You lost to bongcloud?  You deserve whatever low rating you now have.

29th November 2008, 08:01am
#8
by JG27Pyth
NYC United States
Member Since: Mar 2008
Member Points: 1467
Zenchess wrote:

I didn't 'start losing'.  I lost a single game to lenny bongcloud.


Losing to Lenny is an incredible badge of honor. I salute you. Lenny and Timmaylivinalie are heroes to us all. I don't even hope to ever play chess on their level.

29th November 2008, 08:07am
#9
by JG27Pyth
NYC United States
Member Since: Mar 2008
Member Points: 1467
erik wrote:

with a high RD your rating can go up as quickly as it an go down :) just get out there and win a few! :)

seriously though, i appreciate the thoughts. i don't think we'll change it now, but good to hear an opinion on the topic.


As long as you're open to comment -- I still think a rating should be provisional until a player has lost at least 5 games and won at least 5 games -- this "provisional status" would have no impact on the "provisional" player, but would drastically lessen the rating impact of victory or defeat for the non-provisional player.

29th November 2008, 08:32am
#10
by Loomis
Durham, NC United States
Member Since: Oct 2007
Member Points: 3084

Zenchess, get back out there and play some games. Your rating will take care of itself.

30th November 2008, 06:58pm
#11
by k05
vancouver Canada
Member Since: Oct 2008
Member Points: 608

u may have lost to lenny,#, in regular chess rules... but no one's king got to the other side so isn't that a draw in bongcloud rules?

30th November 2008, 07:05pm
#12
by Beelzebub666
Great Britain
Member Since: Feb 2008
Member Points: 378
JG27Pyth wrote:
erik wrote:

with a high RD your rating can go up as quickly as it an go down :) just get out there and win a few! :)

seriously though, i appreciate the thoughts. i don't think we'll change it now, but good to hear an opinion on the topic.


As long as you're open to comment -- I still think a rating should be provisional until a player has lost at least 5 games and won at least 5 games -- this "provisional status" would have no impact on the "provisional" player, but would drastically lessen the rating impact of victory or defeat for the non-provisional player.


That's pretty much what the RD rating does, only it's a smooth transition rather than a sudden change after x number of games.

30th November 2008, 07:14pm
#13
by xMenace
Rothesay, NB Canada
Member Since: Aug 2008
Member Points: 846

Lost me at paragraph 8, but welcome back.

30th November 2008, 07:35pm
#14
by likesforests
United States
Member Since: May 2007
Member Points: 4407

Zenchess, first, I agree you're a wee bit stronger than 1528. Tongue out

It's one of the strengths of the Glicko system that RD increases when you haven't played in awhile--in this case, you acknowledge you became stronger after your absence, and Glicko measures such a change faster than Elo.

It seems like what went wrong, applying this to correspondence chess, is your game was decided pre-vacation but calculated using your post-vacation RD. Maybe rating change should be based on RD(avg) of all your moves rather than RD(final)?

Oh well. I bet you'll be 2000 in a few weeks. :)

 

Add your comment:

Join Chess.com for free to add your comment! Already a member? Then login now to comment.