Re: ECAC Byes and Home-Ice 2011-12
I think this is something that I'm going to add to my run-downs.
Here are the probabilities of team's finishing in each of the twelve places if a team's likelihood of winning is based upon current KRACH.
There are multiple caveats with this.
1) KRACH is a retrodictive, not predictive, system.
2) I used a straight tie percentage of 13.0% (approximately the rate of ties in men's college hockey over the past few years).
3) A team's likelihood of winning is the normal win percentage (KRACH(1) / [KRACH(1) + KRACH(2)]) minus half of the tie percentage.
4) There were only 2500 trials run (which, really, is a pretty low number).
5) KRACH is not updated with the completion of a game (so if, for example, Brown were to win their first three games, one would assume that their KRACH would go up and they may even be favored in their game against Princeton).
Some of these caveats can be seen in the data. A percentage of 0.0 means that that finish occurred exactly once in 2500 trials. A percentage of "xxxx" means that that finish can happen, but did not occur.
One final note is that this is sorted by current standings, not projected standings (like lugnut's predictions are).
I think this is something that I'm going to add to my run-downs.
Here are the probabilities of team's finishing in each of the twelve places if a team's likelihood of winning is based upon current KRACH.
There are multiple caveats with this.
1) KRACH is a retrodictive, not predictive, system.
2) I used a straight tie percentage of 13.0% (approximately the rate of ties in men's college hockey over the past few years).
3) A team's likelihood of winning is the normal win percentage (KRACH(1) / [KRACH(1) + KRACH(2)]) minus half of the tie percentage.
4) There were only 2500 trials run (which, really, is a pretty low number).
5) KRACH is not updated with the completion of a game (so if, for example, Brown were to win their first three games, one would assume that their KRACH would go up and they may even be favored in their game against Princeton).
Some of these caveats can be seen in the data. A percentage of 0.0 means that that finish occurred exactly once in 2500 trials. A percentage of "xxxx" means that that finish can happen, but did not occur.
One final note is that this is sorted by current standings, not projected standings (like lugnut's predictions are).
Code:
----------------------------------------------------------------------------------------------------
| [B]Team[/B] | [B]1st[/B] | [B]2nd[/B] | [B]3rd[/B] | [B]4th[/B] | [B]5th[/B] | [B]6th[/B] | [B]7th[/B] | [B]8th[/B] | [B]9th[/B] | [B]10th[/B] | [B]11th[/B] | [B]12th[/B] |
----------------------------------------------------------------------------------------------------
| Union | 56.8 | 29.1 | 13.6 | 0.3 | 0.1 | | | | | | | |
| Cornell | 34.3 | 43.8 | 20.3 | 1.6 | 0.0 | xxxx | xxxx | | | | | |
| Colgate | 8.8 | 26.1 | 49.7 | 14.1 | 1.1 | 0.2 | xxxx | | | | | |
| Harvard | xxxx | 0.9 | 15.0 | 62.0 | 14.0 | 5.5 | 2.2 | 0.4 | xxxx | xxxx | xxxx | |
| Clarkson | xxxx | 0.0 | 0.6 | 10.2 | 22.4 | 39.2 | 18.4 | 6.6 | 2.2 | 0.2 | xxxx | |
| Quinnipiac | | xxxx | 0.6 | 9.5 | 50.1 | 23.4 | 11.1 | 4.1 | 1.0 | 0.2 | xxxx | xxxx |
| St. Lawrence | | xxxx | 0.0 | 1.7 | 7.3 | 15.5 | 35.3 | 25.1 | 11.7 | 3.2 | 0.2 | xxxx |
| Yale | | | | 0.3 | 1.8 | 5.2 | 12.3 | 28.7 | 27.4 | 17.4 | 6.2 | 0.8 |
| Dartmouth | | | xxxx | 0.2 | 3.2 | 10.2 | 18.4 | 22.7 | 26.4 | 13.2 | 4.5 | 1.2 |
| Princeton | | | | xxxx | 0.0 | 0.6 | 1.7 | 8.4 | 19.7 | 37.0 | 18.8 | 13.8 |
| Rensselaer | | | | xxxx | xxxx | 0.1 | 0.2 | 2.0 | 5.7 | 16.0 | 43.7 | 32.3 |
| Brown | | | | | xxxx | 0.2 | 0.4 | 2.0 | 5.9 | 12.9 | 26.6 | 51.9 |
----------------------------------------------------------------------------------------------------
Last edited: