Tuesday, February 1, 2011

Predicting the First Round of GL Region Rankings

The first set of regional rankings are due out tomorrow, but let's see if we can crack the code a day early.

I'll start off by limiting my potential list of teams to those who have won at least 70% of their regional games. This isn't an NCAA rule or anything, but I think it will pear down our list to a manageable number. It's a shortcut, but I won't tell if you won't tell.

This leaves us with a list of ten:

Winning Percentage (in-region games only)
1. Wooster -- 1.00
2. Hope -- 0.909
3. Penn State-Behrend -- 0.889
4. Marietta -- 0.875
5. Wabash -- 0.833
6. Wittenberg -- 0.800
7. Thiel -- 0.769
8. Bethany -- 0.737
9. Capital -- 0.706
10. Calvin -- 0.700

If we keep the above list, we can compare these teams based on the other Primary criteria. We'll rank them based on their position in the region (41 teams).

Strength of Schedule
7. Hope -- 0.543
16. Thiel -- 0.512
19. Wabash -- 0.506
21. Calvin -- 0.504
22. Wooster -- 0.504
25. Bethany -- 0.497
29. Capital -- 0.484
34. Wittenberg -- 0.469
39. Marietta -- 0.442
41. Penn State-Behrend -- 0.405

I like to start rankings with a mathematical combination of the winning percentage and strength of schedule. This gives me a base line for head-to-head and results versus common opponent adjustments. Now, if we averaged the ranking of winning percentage and strength of schedule, we get the following list:

(More after the jump)


4.5 -- Hope
11.5 -- Wooster
11.5 -- Thiel
12 -- Wabash
15.5 -- Calvin
16.5 -- Bethany
19 -- Capital
20 -- Wittenberg
21.5 -- Marietta
22 -- Penn State-Behrend

But this probably isn't a good way to rank teams. The SOS numbers get pretty bunched up in the middle, and there may not be all that big of a difference between some of the ranking positions. So, I like to employ an RPI-type calculation.

I've been releasing weekly RPI rankings on Mondays (as you've likely seen), and in those rankings, I use the standard RPI calculation WP x 0.25 + SOS x 0.75.

The standard RPI calc yields the following ranking (and RPI values):

1. Hope -- 0.636
2. Wooster -- 0.628
5. Wabash -- 0.588
7. Thiel -- 0.576
8. Bethany -- 0.557
9. Calvin -- 0.553
12. Wittenberg -- 0.552
13. Marietta -- 0.551
15. Capital -- 0.540
17. Penn State-Behrend -- 0.526

We see a lot of teams with lower winning percentages but high strength of schedules sneak into our rankings. I didn't list them, but you can see where they would place.

So let's try a different combination. Maybe a 50-50 winning percentage and strength of schedule split (0.5 x WP + 0.5 x SOS). (I've called this number RPI50 in the past). This calculation yields the following results:

1. Wooster -- 0.752
2. Hope -- 0.727
3. Wabash -- 0.670
4. Marietta -- 0.659
5. Penn State-Behrend -- 0.647
6. Thiel -- 0.640
7. Wittenberg -- 0.634
8. Bethany -- 0.617
10. Calvin -- 0.602
12. Capital -- 0.595

So, Wooster, Hope, and Wabash all appear pretty safe to be ranked. The regional committee will rank six Great Lakes teams, and these three are in the top six in both of these RPI-style calculations.

Marietta, Theil, and PSU-Behrend appear to be candidates to fill out the last three spots; they're ranked in the top six in one of the methods.

Head-to-Head
Wooster beat Bethany, Wabash, and Wittenberg.
Hope beat Calvin and lost to Calvin.
Penn State-Behrend has not played any of these ten teams.
Marietta lost to Capital.
Wabash lost to Wooster and Wittenberg.
Wittenberg beat Wabash and lost to Wooster.
Thiel beat Bethany.
Bethany lost to Wooster and Thiel.
Capital beat Wittenberg and Marietta.
Calvin beat Hope and lost to Hope.

Versus Regionally Ranked Opponents
It's difficult to do this adjustment without the benefit of even one prior ranking, and we can't rightly guess who would be ranked, and adjust accordingly, without first figuring out the rankings anyway, can we? We'll skip this step for now and just pay special attention to the head-to-head category.

Common Opponents
We'll use this to break a tie, if necessary.

My Rankings
I like Wooster as my #1 team. They're the only undefeated team in the region, their strength of schedule is high enough that it shouldn't really ding them, and they have a win against Wabash, who should also end up in our top six.

Hope should end up #2 in the region. They fell in the top two of both of the RPI calcs above. They don't have any games against "ranked" teams (I don't expect Calvin to be ranked this week), and consequently no wins, but I'm not sure they'll be hurt by that fact. Their top-flight WP and SOS should do just fine.

It gets really tricky from here on out, but I think Wabash will end up in the #3 slot. You could argue that they shouldn't be above Wittenberg, based on the head-to-head result, but Wabash has a better win percentage, a stronger schedule, and a nice looking win over Randolph-Macon, who will end up being ranked in the South Region. I just don't think the one head-to-head game overcomes the difference. They have two common opponents with different results. Wabash beat Otterbein but lost to Ohio Wesleyan, and Wittenberg beat Ohio Wesleyan but lost to Otterbein. Let's call that a push.

Marietta is my choice for the #4 position. I'm down to considering Marietta, PSU-Behrend, Thiel, and Wittenberg. I wouldn't be overly surprised to see any of these teams in any order, but when it comes down to it, Marietta has the best winning percentage of the group (except PSU-Behrend, but their SOS is so terrible that they shouldn't count).

I'll say Penn State-Behrend will end up #5. Not because that's where I think they should be, but because that's where I think they will be. The NCAA seems to love win percentage above all else, even if it comes with a bad SOS. PSU-Behrend will get love because of their 0.889 win percentage. But they're not a completely unjustifiable pick; they do have a win over La Roche, who will probably be ranked in the Mid-Atlantic Region. I don't really feel strongly about PSU-B. I'd rather see them #6 or unranked, but my gut says that they'll be higher than that.

I'm predicting that Thiel will gain the #6 spot. I'd like to rank Wittenberg somewhere, but I don't see them passing Thiel, and, in the end, having three NCAC teams in the top six seems like too many. Thiel ranked ahead of Witt in both of our above RPI-style calculations, and they're 1-0 versus (likely) regionally ranked teams (compared to Witt's 1-1 record).

We'll see how this all turns out in approximately 24 hours.