3/2 Regional Rankings

[elbojpb]

Ouch

Error! The game against the ECAC E are conference games, which totally balance out in calculating the OWP They may have gotten to play UNE, but they also had to play Norwich. League games balance out when you play every member of the league the same number of times.

Darn - I've never gotten my hand slapped in red before. Your mother a nun?

I don't dispute your assertion, I simply believe that the probability of 6 of the top 11 teams coming from the same league ~cannot~ be realistic. With the notable exception of NU, the ECAC/E has had a less than stellar season; only one team with a winning conference record.

You can 'splain it anyway you wish, and I understand your defense of the East, but 6 of the best 11 teams coming from any one conference is {in my mind} dubious. And if I were the NESCAC, there's no way in Hades i'd ever vote to disband the interlock.

[spwood]

Re: 3/2 Regional Rankings

Error! The game against the ECAC E are conference games, which totally balance out in calculating the OWP They may have gotten to play UNE, but they also had to play Norwich. League games balance out when you play every member of the league the same number of times.

That would be true if they were listed as one conference. However, the NESCAC would be the only conference in the country to have an over .500 conference mark, right? All of the other conferences, by definition, have to be .500, while the ECAC-E is obviously well below .500. That doesn't have any effect on the numbers? I'm not sure that makes sense to me....

[NUProf]

Re: 3/2 Regional Rankings

That would be true if they were listed as one conference. However, the NESCAC would be the only conference in the country to have an over .500 conference mark, right? All of the other conferences, by definition, have to be .500, while the ECAC-E is obviously well below .500. That doesn't have any effect on the numbers? I'm not sure that makes sense to me....

Look at the pool of teams - there are 20 teams, 19 games per team. There are a total of 190 league games spread out in that pool. There are 190 wins that will be awarded if we count a tie as a half a win for each team. If you add up the league wins for all teams in the NESCAC and ECAC East, that's what you get.

Your OWP can be calculated for your league games in the following way - Deduct your own games. You have 19 opponents who have each played 18 games. Each of those games has a winner (in your schedule) and a loser (also in your schedule), or the two teams tied. There are 171 games played, which means that a total of 171 wins. If you add up the win loss record of each team, there are 19 games per team with 18 teams that makes 342. You calculate winning percentage by your opponents total wins divided by total games played. 171/342 = 0.500. If it were not a complete interlock, these numbers would work out differently. This does not account for the NC schedule played by each team. This same calculation (or a variation therof) can be applied for the league portion of every league in which the number of times each team plays each other is the same. In a league with an unbalanced schedule such as the MCHA, this is not a true statement

I'm sure the math razzle-dazzle lost some people - but hey math is what I do

[spwood]

Re: 3/2 Regional Rankings

I'm sure the math razzle-dazzle lost some people - but hey math is what I do

Well, math is what I do too, so I'm with you. I just couldn't get my head around it before. Thank you for clearing it up!

[one_to7]

Re: 3/2 Regional Rankings

Look at the pool of teams - there are 20 teams, 19 games per team. There are a total of 190 league games spread out in that pool. There are 190 wins that will be awarded if we count a tie as a half a win for each team. If you add up the league wins for all teams in the NESCAC and ECAC East, that's what you get.

Your OWP can be calculated for your league games in the following way - Deduct your own games. You have 19 opponents who have each played 18 games. Each of those games has a winner (in your schedule) and a loser (also in your schedule), or the two teams tied. There are 171 games played, which means that a total of 171 wins. If you add up the win loss record of each team, there are 19 games per team with 18 teams that makes 342. You calculate winning percentage by your opponents total wins divided by total games played. 171/342 = 0.500. If it were not a complete interlock, these numbers would work out differently. This does not account for the NC schedule played by each team. This same calculation (or a variation therof) can be applied for the league portion of every league in which the number of times each team plays each other is the same. In a league with an unbalanced schedule such as the MCHA, this is not a true statement

I'm sure the math razzle-dazzle lost some people - but hey math is what I do

It will all even out for strength of schedule purposes, but it does allow a lot of teams in the nescac to pile up theier win % based on the ecac-e going under .500 in league. Add in some both the ecac-e and nescac fluffing up their overall win % by beating some not-so-good mascac or ecac-ne teams, and you end up with owp and oop records that at worst hover around .500 and at best get bumped well above. All while maintaining a decent win %.

All it takes is basically going .500 in conference games and 4-0 against the ecac-ne and you will be a ranked team. That isn't exactly the definition I'd pick for 'ranking' a team. Personally I've felt that the top 10 eastern and top 6 western teams would make for more deserving rankings.

All you realistically need for rankings are however many autobids are in your region + however many at large bids there are. Granted that could leave some tough competition out of consideration for being ranked, but ranking 22 overall teams is a bit much, in my opinion.

end of rambling rant.

[NUProf]

Re: 3/2 Regional Rankings

It will all even out for strength of schedule purposes, but it does allow a lot of teams in the nescac to pile up theier win % based on the ecac-e going under .500 in league. Add in some both the ecac-e and nescac fluffing up their overall win % by beating some not-so-good mascac or ecac-ne teams, and you end up with owp and oop records that at worst hover around .500 and at best get bumped well above. All while maintaining a decent win %.

All it takes is basically going .500 in conference games and 4-0 against the ecac-ne and you will be a ranked team. That isn't exactly the definition I'd pick for 'ranking' a team. Personally I've felt that the top 10 eastern and top 6 western teams would make for more deserving rankings.

All you realistically need for rankings are however many autobids are in your region + however many at large bids there are. Granted that could leave some tough competition out of consideration for being ranked, but ranking 22 overall teams is a bit much, in my opinion.

end of rambling rant.

NCAA guidlines say that approximately 1/3 of the teams in each region will be ranked. Again, that's one of those "one size fits all sports" rules that the NCAA is so good at creating.

[Dyce]

Re: 3/2 Regional Rankings

NCAA guidlines say that approximately 1/3 of the teams in each region will be ranked. Again, that's one of those "one size fits all sports" rules that the NCAA is so good at creating.

I'm guessing part of it has to do with creating large enough sample sizes for the "record vs. ranked teams" criterion to be anything resembling meaningful. The NCAA knows full-well that the bottom of the list is irrelevant for Pool C purposes, but they want to provide some sort of clear index of "good" teams they can use for the sake of comparison.

[NUProf]

Re: 3/2 Regional Rankings

I'm guessing part of it has to do with creating large enough sample sizes for the "record vs. ranked teams" criterion to be anything resembling meaningful. The NCAA knows full-well that the bottom of the list is irrelevant for Pool C purposes, but they want to provide some sort of clear index of "good" teams they can use for the sake of comparison.

That would be the reason. Perhaps for a more stable system, they could go to 1/4 of the teams, but then use what they did in the past which is use the once ranked, always ranked process that they used in the past. The current system changes a team's profile when a team that was formerly ranked becomes unranked - and they didn't do a darn thing. If they beat that team they are hurt, if they lost to them they are helped. Once ranked always ranked made more sense from a statistical point of view.