Boylston Chess Club
I just patented pi.
I don't understand the point of giving a bonus with black? If two teams are equal in all ways except one team is better with black, wouldn't that mean that the other team is better with white? Since each match involves two whites and two blacks, it seems like a completely arbitrary to put emphasis on one while completely ignoring the other?
The black piece bonus is based on the idea that all things being equal, among good players it is harder to win with Black than with White. However, your point has some validity also.
I expect that I will tweak the methodology as I get more data and learn how it works in practice. If this variable and/or its weighting is having a disproportionate effect on the rankings, I may make some adjustments.
Since we have data from the previous two seasons it should be possible to tweak the model parameters. Most importantly, does the power ranking behave as you would like over the course of the previous two seasons?
Nice! Maybe the USCL should have its own BCS computer that everyone hates.
BTW, I just love when people complain about the BCS computer rankings, because after all it just ranks 'em according to the rules that we give the computer. If the computer is getting it wrong, then give the computer different parameters that fit with what we like.
Oh-- but this is because the human polls take into account important subjective things, like how much money will certain matchups generate, how much do particular coaches complain about the negative or positive publicity, and how popular are certain schools. The computer doesn't care about those things, so we hobble the computer with our subjective rankings.
DG, it just can't be that the Knights are at the bottome of the power rankings. This is the team from NEW YORK, and everyone LOVES the New York teams (or loves to hate). That's got to count for something, if we follow the sports models-- maybe put Tennessee at the bottom and New York 11, just for the visibility fudge factor alone.
BDK - Interesting idea, but way too much work. I already have one full-time blogging job. 
If you are so motivated, I'll send you my model and you can try it. If it actually has predictive power, don't tell anyone but me. Together I'm sure we can make a killing at Pinnacle.
BDK - Never mind. I just went to Pinnacle and it loks like they are not taking bets on the USCL anymore. Maybe you can adapt it to pick stocks?
Finally, a list where the Tempo aren't last!!!
Looks like Shmelov - Critelli will be the key matchup. Board 1 could go either way and Kleiman and Krasik have uphill battles, to say the least.
Where's Mark Glickman when you need him?
I'd be happy to take a look and plot the rankings as a function of week for the past two seasons. I'd leave the analysis of the answers, and whether they look reasonable, to the peanut gallery since I haven't followed the league and I'm just a patzer anyway. You'd have to let me know where I can find that data (I assume at their web site).
I bet principal components analysis would pull out a single number that correlated very highly with performance (this is basically what the IQ is: we could have a chess version). I'd compare the DG method with principal components.
hey this is chris i wanna wish my team good luck and im sorry to david for my previous comments, thnks.
There was an article in Scientific American that pointed out the fallacy of such attempts at ranking things made up of multiple variables like teams, dice, and the democracy paradox which is related. The jist of it is that it is mathematically possibly for A to be a statistical favorite over B, while B is better than C, yet C trumps A.
So no computers or humans can actually solve this problem. By the way, there is a related problem in making lineups - or choosing a chess opening. There is no correct choice, but there is a correct way to decide.
I'll leave you to ponder that. Apparently you got the time 
By the way WTF is up with the Newt Gingrich SPAM!!? Gag me in a restroom.
smokinoak: ah the rock scissors paper paradox.
Rock beats scissors, scissors beats paper, therefore rock beats paper.
Hey wait I screwed that up:
Scissors beat paper, paper beats rock, therefore scissors beat rock.
Or
Paper beats rock, rock beats scissors, therefore paper beats scissors.
Hmm, in my first post it seems I managed to pick the only one that doesn't work.
I'm all for as many different rankings of teams as possible - there's nothing for a league like debating if Team A could beat Team B, and why commentators A through Z are all off their rockers.
In case anyone went to take a look at my rankings tonight and didn't see them for a few hours, there's an explanation up there now. I'll be back to blogging tomorrow.
Orange King we look forward to future entries in your USCL blog. Please email me at admin@uschessleague.com whenever you update it, so that we have the option of putting it on the front page as quickly as possible.
You had some interesting stuff up there before it all dissapeared.
Commenting by HaloScan
