PDA

View Full Version : How Early is Too Early? Inside the PWR



Pages : [1] 2 3

Priceless
01-07-2010, 01:23 PM
How early is too early? It might seem ridiculous to look at the PWR now, before the conference schedule really heats up, but history tells another story. Over 75% of teams that are in the field at New Year's make the tournament in March. Of the 97 teams that qualified in the PWR, 73 were in the field in early January (75.26%). Among the top 4 (number 1 seeds) 27 of 28 teams qualified. Only the 2004 St. Cloud squad fell from those heights. That's great news for Miami, Denver, Colorado College and Ferris State.

The news is almost as good for Wisconsin, Bemidji, Michigan State and Boston College. Of the January #2 seeds, the success rate is 20/28 (71.4%) with only Colorado College being a repeat offender. The Badgers should be a little nervous though. Of the eight teams that missed the playoffs, 50% were from the WCHA.

There is a slight dropoff among the #3 seeds, as only 16 teams (57%) from January made the field. Vermont is sitting at #9 in the current PWR, but the Catamounts were ranked #9 and #10 in 2006 and 2007 and missed the playoffs. Duluth and St. Cloud have dropped out in the past as well.

The last few teams in are extremely volatile, for one reason because we're not sure where the cutoff will be. In the seven seasons examined there were only 15 spots that weren't reserved for autobids. But of those 15 teams, nine (60%) were in the field in January.

If your team is close, they still have a chance. Almost a quarter of the field were not included in the January field. Last year Yale wasn't even in the January PWR (#26 RPI) and wound up a #2 seed. North Dakota was #23 in the PWR. In 2007, BU and UMass were in the 20's in January and still qualified. In the January 2006 PWR Michigan State, BU, Maine and UNH were ranked 19-22 and all made the tourney.

The New Year's PWR is also fairly good at showing us where teams will be seeded. 30% of teams in the field were within their bands in the January rankings. There were even 10 matches of 97 (10.3%) that were in the same place in the PWR that they were in when the final PWR came out.

Here are the current PWR

1 Miami
2 Denver
3 Ferris State
4 Colorado College
---
5 Wisconsin
6t Bemidji State
6t Michigan State
8 Boston College
---
9t Vermont
9t St. Cloud State
11 Minnesota-Duluth
12t Yale
---
12t Massachusetts
12t North Dakota
15t Union
15t Maine
17t New Hampshire
17t Mass.-Lowell
19 Lake Superior
20 Alaska
21t Quinnipiac
21t Cornell
23t Northern Michigan
23t Minnesota State
25 Notre Dame

Miami, Denver, Ferris State and CC are not only locks for the tournament, they are almost 60% favorites to garner #1 seeds come March.



For this analysis, the PWR from 2003-2009 was used. Dates examined were New Year's, the second weekend in January, post-Beanpot, the end of February and at the beginning of the conference tournaments. The second report will follow in two weeks with analysis of that weekend's PWR compared to the past. The PWR does become a more accurate predictor as the season progresses.

To learn more about the PWR, visit the USCHO PWR page and the Do-It-Yourself Calculator at http://www.slack.net/~whelan/tbrw/tbrw.cgi?2010/rankings.diy.shtml

Hammer
01-07-2010, 01:26 PM
We'll know a ton more about Ferris in about 17 days. They've got Notre Dame, Miami and Michigan in the next three weeks before the schedule gets noticeably lighter.

ExileOnDaytonStreet
01-07-2010, 01:30 PM
Pretty solid look at the predicting power of the PWR:

http://blog.siouxsports.com/2009/12/31/when-to-start-looking-at-pwr/

The second graph is the most telling.

jcarter7669
01-07-2010, 01:34 PM
Nice analysis, it will be interesting to see what things look like in another few weeks.

5mn_Major
01-07-2010, 01:54 PM
Def not too early for the PWR IMO. Frankly...its accurate in November by its system of measurement. Its its misinterpretation that is only the issue.

Whether its too early for bracketology based on the PWR...doesn't matter, if folks like to discuss...thats what message boards are for.

(just addressing the question :) )

Goon
01-07-2010, 01:55 PM
I agree not to early to be looking at the number and seeing how they might play out. Looks like there could be 5-6 WCHA teams this season in the NCAA tourney.

Priceless
01-07-2010, 01:58 PM
Def not too early for the PWR IMO. Frankly...its accurate in November by its system of measurement. Its its misinterpretation that is only the issue.

Whether its too early for bracketology based on the PWR...doesn't matter, if folks like to discuss thats what message boards are for.

(just addressing the question :) )

It's probably too early for serious bracketology, but it's still fun to look at where things are now and pretend you're on the committee trying to balance bracket integrity with attendance, etc. But the PWR are very accurate even this far out at telling us who will be in the eventual field.

Federal League
01-07-2010, 02:02 PM
Look at all the HE teams on the bubble. UMass, Maine, UNH and Lowell all in the 12-17 range. And BC and Vermont certainly haven't locked anything up yet. Gonna be interesting to see which of those teams can do what it takes to get in.

candycanes99
01-07-2010, 02:20 PM
How early is too early?

Anything before March 1st, 2010.

'cuz I ain't ready yet. :p

LynahFan
01-07-2010, 02:23 PM
I agree not to early to be looking at the number and seeing how they might play out. Looks like there could be 5-6 WCHA teams this season in the NCAA tourney.
But don't forget Priceless's note that most of the teams that drop out from Jan to March are from the WCHA. That's not random - after beating up on weaker non-conference foes (yes, I'm man enough to admit it), the WCHA's mid-season ratings are high, but they will necessarily drop off a bit as they beat each other up and trend back toward a collective .500 record.

blockski
01-07-2010, 02:32 PM
For this analysis, the PWR from 2003-2009 was used. Dates examined were New Year's, the second weekend in January, post-Beanpot, the end of February and at the beginning of the conference tournaments. The second report will follow in two weeks with analysis of that weekend's PWR compared to the past. The PWR does become a more accurate predictor as the season progresses.

The one thing I'd note is the 10 game minimum for TUC to count. There are some teams ranked quite highly (Ferris State, Bemidji) that do not yet have anywhere close to 10 total games against TUCs. That will certainly change for Ferris as they get deeper into their CCHA season, and when that comparsion clicks in and counts, they might see a larger adjustment than otherwise expected.

Cat lover
01-07-2010, 02:35 PM
But don't forget Priceless's note that most of the teams that drop out from Jan to March are from the WCHA. That's not random - after beating up on weaker non-conference foes (yes, I'm man enough to admit it), the WCHA's mid-season ratings are high, but they will necessarily drop off a bit as they beat each other up and trend back toward a collective .500 record.


As most of the teams in the stronger leagues are done now with OOC games it would seem that for those teams the SOS will not change much from now to the end of the season, what it now comes down to is WINS pretty or ugly its time to "just win baby".

Those teams in weaker divisions really need to win more just to hold the spots they have as they will be the ones to see the SOS drop.

Hammer
01-07-2010, 02:38 PM
The one thing I'd note is the 10 game minimum for TUC to count. There are some teams ranked quite highly (Ferris State, Bemidji) that do not yet have anywhere close to 10 total games against TUCs. That will certainly change for Ferris as they get deeper into their CCHA season, and when that comparsion clicks in and counts, they might see a larger adjustment than otherwise expected.

I wrote about that last night in the Ferris thread. If we take Ferris' TUC record right now and use it, Ferris would drop 6 comparisons, dropping them down to #9 in the PWR. That's why the next 3 weeks are absolutely crucial for us and our tournament chances.

Rich
01-07-2010, 02:49 PM
Great analysis - thanks!

On a similar but slightly different note, I find the following useful to see what my team will need to do for the rest of the season to make the tournament. The PWR generally follows the RPI, and Sioux Sports has a great RPI prediction section. A decent guess is that a minimum of 0.5500 is needed to make the top 14 and have a shot at making the tournament. You can see what your team would need to do to achieve a 0.5500 RPI. Using linear interpolation, you can see from the following link that CC would need to go 8-8 during their remaining regular season games to achieve a 0.5502 RPI. 10-6 should put them in the top 8, and 11-5 might be enough for them to remain in the top 4 and earn a #1 seed in a Regional.

Obviously the RPI required to make each of these levels is determined by how the other teams perform, so none of it is perfect. The RPI predictor isn't perfect either. But I've found that the RPI required to make the tournament has been pretty close in the past, and it lets you get some idea of what your team will need to to go have a chance to get in. While it's not perfect, it's much better than a guess.

http://siouxsports.com/hockey/rankings/rpidetails.php?teamid=2

Priceless
01-07-2010, 02:50 PM
The one thing I'd note is the 10 game minimum for TUC to count. There are some teams ranked quite highly (Ferris State, Bemidji) that do not yet have anywhere close to 10 total games against TUCs. That will certainly change for Ferris as they get deeper into their CCHA season, and when that comparsion clicks in and counts, they might see a larger adjustment than otherwise expected.

If you remove the 10-game restriction, Bemidji moves up to #5 and Ferris falls only to #7. They both win a bunch of common opponent comparisons and currently have a high RPI.

blockski
01-07-2010, 02:51 PM
I wrote about that last night in the Ferris thread. If we take Ferris' TUC record right now and use it, Ferris would drop 6 comparisons, dropping them down to #9 in the PWR. That's why the next 3 weeks are absolutely crucial for us and our tournament chances.

Right - the catch is that if they keep that same record against TUCs, their RPI will likely drop some as well, so all of that drop can't be attributed to the TUC comparison clicking in - but it's certainly something to keep an eye on.

Look at Bemidji - they might not get to 10 games. They're at 5 right now, with 2 coming up against UMD. Other than that, they've got two against RPI #26 UNO later in the season, and they've already played RPI #28 Minnesota twice - but they've also played RPI #24 Minnesota State twice (and lost both).

All of those teams could move in or out of BSU's RPI resume, which will both change the chances of them ending the season with at least 10 TUC games, and also potentially changing the pool of results against those TUCs. For example, if Minnesota (with whom BSU split) became a TUC instead of Minnesota State (who swept BSU), that's a net gain for the Beavers - IF they get to 10 games overall...

Hammer
01-07-2010, 02:52 PM
Priceless, unless you've got something else I haven't seen, a rough cut tells me that Ferris' TUC record would cause them to lose the comparisons to Wisconsin, Yale, North Dakota, UMass, Union and Alaska. The MSU comparison would also be extremely close, but Ferris would still hang on to that one at the moment. That moves them from 22 points to 16, sliding them down to #9.

Priceless
01-07-2010, 02:55 PM
Priceless, unless you've got something else I haven't seen, a rough cut tells me that Ferris' TUC record would cause them to lose the comparisons to Wisconsin, Yale, North Dakota, UMass, Union and Alaska. The MSU comparison would also be extremely close, but Ferris would still hang on to that one at the moment.

I used the calculator at the link in my first post. I reduced the TUC requirement to 1 game.

Hammer
01-07-2010, 02:57 PM
That's fine. I did mine by the seat of my pants at 12:30 this morning. :o I just looked at the individual comparisons that Ferris currently carries and went from there.

MaizeRage
01-07-2010, 03:03 PM
Look at Bemidji - they might not get to 10 games. They're at 5 right now, with 2 coming up against UMD. Other than that, they've got two against RPI #26 UNO later in the season, and they've already played RPI #28 Minnesota twice - but they've also played RPI #24 Minnesota State twice (and lost both).

They'd need Northern Michigan, Minnesota, Minnesota State, and Nebraska-Omaha all to end up on the right side of the top 25, or hope that Western Michigan goes on a run and replaces one of them to reach the 10-game minimum, which seems really unlikely to me.