Jan 142014
 

Remember when we posted right before the turn of the year about UAH’s presence in the KRACH ranking?  I looked at it again today because I wanted to create a predictive tool for the remainder of the 2013-14 WCHA regular season schedule.  As a reminder, the WCHA takes eight teams into the playoffs, with the top four team hosting the first round.  That leaves one other school out in the cold with our beloved Chargers.  Who will that team be?  Who will host the first round playoffs?  Will Ferris State remain undefeated in league play?  These are the questions that I sought to answer.

Let me note this: any comparison-based ranking system is rearward-facing.  If UAH knocks off Anchorage twice this weekend, the Chargers would jump in KRACH, probably all the way to 55th.  That’s because KRACH takes a look at what has already happened and does comparisons between common opponents.  I think that most sports fans are familiar with the Ratings Percentage Index (RPI) method of ranking: your rank is affected by your strength-of-schedule (SOS) and your own winning percentage.

But what if we took those comparisons and used them as a sort of forward-looking predictive tool?  To do this, I used the following method:

Expected points = (My KRACH) / (My KRACH + Your KRACH) * 4

I computed these in a raw manner: of the four points available this weekend in Huntsville, this method garners 3.72 to Anchorage (132.5) and just 0.28 to the Chargers (9.861).  This obviously doesn’t work, as you don’t get points for winning a period.  In this comparison, Anchorage can be expected to win both games and go from 7-7-2 to 9-7-2, which jumps them from fifth to fourth in the standings (against all of my other comparison-based predictions for Week 17).

Now, I used my own judgments in making these calls.  For example: Alaska and Northern are 2.13 and 1.87 respectively in expected points this weekend.  You can argue that this should be a split, and I agree, especially because the Wildcats are at home.

Another example: next week, Tech hosts Ferris.  By KRACH, the Bulldogs should pick up 3.31 points.  Should they be given the sweep, or should they just head back to Big Rapids with a win and a tie?  I went with the latter because the Huskies are at home.

My third and final example: Anchorage hosts Northern next weekend, with expected points values of 2.36 and 1.64, respectively.  I gave the Seawolves a win and a tie as a home Alaska team, even though straight rounding would make it a split.

Here’s a summary of the results (playoff hosts, miss playoffs):

Team Record Points
1 Ferris State 20-0-8 48
2 Minnesota State 16-10-2 34
3 Bowling Green 13-9-6 32
4 Alaska-Anchorage 13-11-4 30
5t Lake Superior 13-13-2 28
Northern Michigan 13-13-2 28
7 Bemidji State 11-11-6 28
8 Alaska 11-14-3 25
9 Michigan Tech 9-12-7 25
10 Alabama-Huntsville 1-27-0 2

 

Note: I haven’t read the WCHA handbook to see how that tie would get settled.  I’m going to guess that it would be in head-to-head match ups as the first tiebreaker, but that matchup doesn’t occur until Week 21, where I’m picking a split.  I’m willing to bet that Shane Frederick or Jack Hittinger will be along to explain this to me.  (Would it be total goals?)

Updated: Matt Wellens of USCHO got me the data.  The tiebreakers are: a) head-to-head points, but only if you’ve played four contests against each other, b) higher number of conference wins, c) comparison of winning percentages against all teams above them, team-by-team down through the table.  I’d spend a lot of time looking at this, but I’d rather not do it unless it’s a reality.  It is important to note that conference wins isn’t the first tiebreaker, which is what I’d have thought it would be.

I’ll be updating this exercise every week, as two things are going to be changing: expected points will turn into real points, and those game results (as well as those non-conference tilts left) will affect KRACH.  I’ve got it up on Google Docs if you’re curious to read it week-over-week.  This is the document that I will update early each week.  I will probably make update posts here this week with a couple of tables (mainly, the summary tab of that spreadsheet) and a little commentary about any moves.

I fully expect that real life will have greater variability than these chalk predictions.  I’m curious to see how the predictive tool will hold up over time.  Will the Huskies come to Huntsville for early tee times?  Will Bemidji slide down and miss the playoffs entirely?  (Please say yes.)  We will see.