Monday, October 17, 2011

Football: Power Rankings, Oct. 17

We have a new team atop the Power Rankings this week — and they're not from the OUA.

posted by Rob Pettapiece at 7:00 AM
Labels: , ,

7 comments:

Superfun Happy Slide said...

I hate to appear or reveal myself as Mustang apologist I am but you can't administer a lower ranking, in a poll where rpi is the most concret metric, because the tough part of the Western's schedule was in the first half of the season; critical victories (including point spread) over quality conference opponents is how the rpi is generally calculated.

Perhaps not playing Windsor makes the difference as the season has advanced?

Subjectively, you're list certainly passes the common sense test. However, if the rpi is going to play a part in the ranking you can't just ignore it . . .

. . . just say'n.

Slide

By the way, Go Stangs Go!

10/17/2011 3:07 PM
sager said...

Yes, you can, because good programs are stronger in weeks 6, 7, 8 than in 1, 2, 3.

10/17/2011 7:07 PM
Rob Pettapiece said...

However, if the rpi is going to play a part in the ranking you can't just ignore it

It isn't ignored. It does play a part in the rankings. In fact, it's nearly half the weight for these power rankings (the other almost-half going to a similar ranking system, and the tiny leftover is the fans' preseason predictions).

critical victories (including point spread) over quality conference opponents is how the rpi is generally calculated

RPI doesn't actually consider point spreads at all. Which is why we look at other measures as well.

10/17/2011 9:10 PM
Superfun Happy Slide said...

Rob, I still think that margin of victory is a part of most rpi ratings; it is in NCAA basketball, for selecting their sixty-four tourney contestants.

Sager, I would readily agree that subjective rankings should take into consideration the swings of momentum (items that contribute to that flow, like injury, ineffectiveness, and the mysterious Roger Clemen's 'second wind.') that a team rides and endures during the season. However, that type of ranking is more of a subjective interpretation, seemingly un-linked to a statistical metric. I think the ranking is right but I fail to understand how a poll that measures head-to-head competition could possibly place the number-2 team in the OUA over the unbeaten leader. Being that they competed, in the Hammer; number-2 lost.

10/18/2011 6:22 PM
Rob Pettapiece said...

Sorry, but you're wrong. Unless the NCAA has recently introduced a brand-new, super-secret RPI that somehow includes margin of victory, and they told absolutely nobody about it except you, there is no RPI I have ever heard of that distinguishes between a 20-point and a 2-point win.

If you don't believe me, how about Ken Pomeroy?

**
How does the RPI factor in margin of victory?
It doesn’t.
**

I really don't know how much clearer I can put it. There may be other rankings that consider margin of victory (the coaches' poll, probably, takes it into account implicitly or explicitly), but the RPI does not.

10/18/2011 8:41 PM
Rob Pettapiece said...

I fail to understand how a poll that measures head-to-head competition could possibly place the number-2 team in the OUA over the unbeaten leader

Assuming you are actually curious about this, and not just upset that your favourite team isn't ranked as high as you would like, I can explain.

1. The difference between McMaster and Western in the overall power rankings is very small. Essentially, Mac is treated as a 7.16-win team (over the course of a full OUA season), and Western is considered to be a 7.06-win team. So to the extent that the Mustangs are "behind", it is by a tenth of a win.

2. Western is higher in the RPI, mostly because they beat Mac, but again, only by about 0.009 RPI points, and the margin of error has to be at least that much. Western might be higher if they hadn't played a weak schedule lately -- the RPI variant we use places more weight on recent games -- but it's not like Mac just played Laval and Calgary, either.

3. Mac is higher in SRS (Simple Ranking System), by two points per game. This is a bigger difference, though likely not significant given the short schedule. Still, why are they higher? Let's look at common opponents:

- Waterloo: Western by 64, Mac by 26. I cut off blowouts after a certain point so there's no extra credit for Western's piling on.

- Laurier: Western by 6, Mac by 19.

- Guelph: Western by 3, Mac by 24.

- Toronto: Western by 12, Mac by 36.

Now does it make sense? Including margin of victory actually makes the 'Stangs look worse.

**

Your objections, if they aren't purple-coloured, boil down to "A beats B so A should be higher than B." I hope you realize that we can't just rely on that rule for all 26 teams, which is why we use more sophisticated measures to understand how teams are performing.

10/18/2011 9:11 PM
Superfun Happy Slide said...

Rob, Valid points; my interpretation of the rpi was based on a CBS, NCAA tournament studio show from the late 90s. I'm positive that they made the point that margins where a part of the rpi. Perhaps they meant it was simply a part of the overall formula. Maybe the Keith Van Horn hoopla (he of Utah and 'white bros can jump' hype) distorted my memory.

Although, not to be overly argumentative, one could argue that using metrics designed to compensate for opponents who do not meet in the regular season is flawed when ranking those who did compete directly with one another. In that sense, it seems irrational that head-to-head match-ups should matter less than said formula; transitivity rarely being trumped in real-world comparisons.

In any event, polls are points of engagement for fan bases and media debate. While my purple-bias will always show through, I appreciate the explaination and rationale behind the poling.

PS. That's what she said.

10/18/2011 9:58 PM

Post a Comment

Subscribe to: Post Comments (Atom)