If you're curious how ELO ratings are calculated, look on wikipedia...basically, I estimated them with 200 runs through, giving everyone a 2000 rating to start. They don't seem to change appreciably after that, so here are the ratings and rankings. (c) = captain, 1st#=relative rating of draftee, 2nd#=order draftee was picked.
(c) suskie 2365
1 bbobb 2212 7
(c) boo 2147
(c) janus 2134
(c) emp 2121
2 venter 2093 6
3 zig 2092 3
4 zipp 2058 1
5 doi 2051 12
6 woodhouse 2047 4
7 true 2039 8
(c) od 2032
8 espiga 2003 13
(c) felix 1975
9 belisarios 1966 11
10 wolfqueen 1955 2
11 aschultz 1948 9
12 sashanan 1940 15
13 randxian 1901 16
14 de 1871 5
(c) dagoss 1865
15 turducken 1842 17
16 disco 1813 10
(c) will 1802
17 vortex 1729 14
People with better records may have worse ratings because they played weaker schedules. If you're wondering about relative strengths, a difference of 120 means an expected 2-1 win. 180 = 3-1(if 4 judges) and 240=4-1, 280=5-1. Basically the ratio = 10^(rating difference - 400).
Ratings are still pretty volatile as I started with 2080 or so, Wolfqueen with 1850, before this round. That changed, obviously! While the sample size isn't enough to be statistically significant, we can see some clear trends.
Anyway I hope people enjoy reading this, as I enjoyed writing a program to calculate this. Don't take it too seriously--rating is not a true indication of worth as a writer, which can't be calculated anyway, and all that.
If anyone has old tourneys and is curious, maybe I could crank out the ratings for them, too.
|Most recent blog posts from Andrew Schultz...|
|jerec - August 10, 2009 (10:22 PM)
Oh yes, that stream of words and numbers made perfect sense. *nods*
I kid. I get the theory of it, and I can see the rankings, so I suppose I don't need to wrap my mind around the mechanics of it.
|JANUS2 - August 10, 2009 (11:27 PM)
I approve of these rankings!
|aschultz - August 11, 2009 (07:45 AM)
Jerec--yeah, it took a bit of time for me to figure out 1) the ratings and 2) a simulation that could estimate what they "should" be.
Janus--I can't say I'm surprised you're high up there. I mean, if we went by votes, you'd be the #3 seed. You guys had that 1-2 5-4 loss to us among other close ones. Switch a vote there and you're in the playoffs.
|randxian - August 11, 2009 (07:45 PM)
Isn't a 1900 or so in chess expert level? So does this mean I'm an expert reviewer?
If so, this ELO rating made me feel worlds better. Thanks man.
|aschultz - August 12, 2009 (09:03 AM)
Randxian--actually, I assumed that 2000 (as a good round number) was the average for us all, and then for my simulation, what I did was to run 200 tourneys in a row with results reflecting what happened in the TT, adjusting the ratings after each tourney. Eventually the ratings stabilize.
I chose 2000 because it was a good round number, but you're right, 2000 would mean expert. It's difficult to see where we'd stack up vs all reviewers, professional or not, but I think it's safe to say HG is well above the usual level of reviewing you'll see on the internet.
So, in the 1900's, we're both a bit below average for the tournament, but we aren't going to be swept by anyone.
|randxian - August 13, 2009 (04:49 PM)
but I think it's safe to say HG is well above the usual level of reviewing you'll see on the internet.
I uh, actually was swept.
|aschultz - August 14, 2009 (07:49 AM)
Oops. I meant, probably should expect not to be swept, because you can expect .5 points mathematically speaking.