April 6, 2007

Conference Standings Odds

West

1.100%
2.76%24%
3.69%31%
4.76%24%
5.34%24%24%18%
6.82%18%
7.69%31%
8.93%6.9%
9.93%6.9%
10.100%
11.97%3.3%
12.90%6.5%3.3%
13.85%8.7%6.5%
14.60%31%8.7%
15.60%40%

East
1.100%
2.100%
3.100%
4.66%34%
5.66%34%
6.87%13%
7.87%13%
8.48%43%8.6%
9.38%33%29%
10.53%23%14%10%
11.63%28%10%
12.73%28%
13.100%
14.100%
15.100%

James Mirtle presented his numbers from good guessing, here are the simulated odds for each position in the standings that makes the playoffs. Each row represents a position in the standings. Across the rows are images and numbers. The numbers say what percent chance that given team (the image) has of being in that position. So Buffalo has a 98% chance of being in first place in the east and N.J has a 1% chance of being in first. Pittsburgh has a 75% chance of being in fifth and New Jersey has a 15% chance of being in fifth, etc.

13 comments:

teebs said...

This would be awesome if I had a clue what that graph was trying to communicate.

JavaGeek said...

I added some better explanations, which I hope help. I hope the little pictures are visible.

PDO said...

Java, mind getting the entire conference listed for those of us stuck hoping for a top 5 pick?

JavaGeek said...

pdo:

I posted the entire table, but Edmonton will likely do worse than predicted, because I haven't rescaled teams based on trade deadline and injury effects.

PDO said...

That was fast.

Thanks Java.

James Mirtle said...

Good stuff. Hopefully this is something you'll update every day.

David Johnson said...

Could you add another column? Since there is 6 teams in the eastern conferecne race you really need 6 columns to present all the information. For example, the Rangers % total only adds up to 62%.

Also, are you actually simulating the individual games or just simulating how many points each team will get?

JavaGeek said...

I'm predicting individual games based on a "pretty good" prediction model. It's important to look at the individual games because points of one team depend on the points of another.

I agree another column would be useful, but you likely can infer the other team if there are 5 listed...

James Mirtle said...

I'm curious what the percentages for each team to make the playoffs are... Atlanta, for example, only shows up in the third spot.

JavaGeek said...

Eastern playoff race

or

Eastern Conference Standings [MP = make playoffs]

David Johnson said...

I agree. Carolina and the Islanders are all but done. Tomorrow's Leaf-Rangers game could be huge. If the Rangers win they will be in very good shape but a loss will put their playoff spot back in play. Anything can happen though because these teams still have quite a few games against each other.

Leafs still play Rangers, Islanders and Canadiens.

Rangers still play Leafs, Islanders and Canadiens.

Montreal still plays Rangers and Leafs.

Should be an interesting final week.

David Johnson said...

You might want to check your algorithm because you have Atlanta with a 0.13% chance of finishing 9th but they have clinched a playoff spot. Toronto could tie them in points and wins but Atlanta would win the tie breaker as they have more points in the season series.

JavaGeek said...

Fixed.

I thought I could be lazy (not include wins tie breaker or GF/GA tie breaker).

All tie breaking rules are now coded.