Updated Brackets - Feb 22nd

Here is my latest bracket projections, with games as of Thursday Feb 21st. If you've seen these before, you know the drill. It's not how I think the brackets WILL look like, it's only how I think the brackets SHOULD look like if the NCAA Tournament Committee actually practices what it preaches.

1: Memphis, North Carolina, Tennessee, UCLA
2: Duke, Kansas, Texas, Wisconsin
3: Connecticut, Washington St, Stanford, Georgetown
4: Butler, Xavier, Indiana, Drake
5: Notre Dame, Michigan St, Marquette, Louisville
6: Purdue, St. Mary's, Vanderbilt, Southern California
7: Pittsburgh, Gonzaga, Arizona, Texas A&M
8: Clemson, Oklahoma, Miami (FL), BYU
9: Kent St, Kansas St, Baylor, Arizona St
10: Oregon, West Virginia, Ohio State, Massachusetts
11: Arkansas, Davidson, California, Stephen F. Austin
12: Rhode Island, Florida, Wake Forest, Mississippi
13: South Alabama, VCU, Oral Roberts, CS Northridge
14: Boise St, Cornell, UMBC, Winthrop
15: Siena, Wagner, Belmont, Portland St
16: American, Lamar, Morgan St, Austin Peay, Alabama St

Last 4 In: Rhode Island, Florida, Wake Forest, Mississippi
First 4 Out: Wright St, Syracuse, UNLV, Illinois St
Next 4 Out: Dayton, Villanova, North Carolina St, Houston

This is probably not a good idea to talk about prior to the Tennessee-Memphis game, but I wanted to look more closely at scheduling and how the RPI SOS numbers are out of whack. Let's compare three teams who we can safely assume have figured out how to beat the system (not passing judgment, I wonder why more programs haven't caught on): Tennessee, Syracuse and Arizona versus one team that is hurting its tourney profile by its non-conference schedule: Washington State.

First, let's look at the RPI SOS and NC SOS of the four teams in question.

Arizona - RPI SOS = 1; NC SOS = 5
Tennessee - RPI SOS = 2; NC SOS = 7
Syracuse - RPI SOS = 13; NC SOS = 19
Washington St - RPI SOS = 58; NC SOS = 227

I'm not going to defend Washington's non-conference schedule, they did play some cupcakes. However, I will contend that despite the numbers above suggesting otherwise, to date:

1. Washington State has played a tougher schedule than both Syracuse and Tennessee.

2. Washington State's cupcake non-conference schedule was not much easier than Syracuse's

If you've taken the time to watch my presentations, you know that the RPI SOS measure is flawed in to two very significant ways. First, the SOS measure is essentially just the average RPI of a team's opponents. So a team playing the 1st, 2nd and 330th ranked team will have the same SOS schedule as a team playing the 110th,111th, and 112th ranked teams, despite the fact that a quality team would likely go 1-2 in the first scenario and 3-0 in the second scenario. That's a biggie. This is why we have 'the bottom feeder effect' which creates the greatest opportunity for programs to game the system. The second isn't quite as big, but it can skew the results. Home vs. road is not accounted for in the RPI SOS number. The wacky +/-40% adjustments that the NCAA put into practice go into the win-loss component and not either of the two SOS components (while also have the extra benefit of doing more harm than good.)

I compute SOS by recognizing that difficulty is not a linear function and difficulty should account for the location of the game. Home-court advantage can be easily quantified (and is not linear either. It matters most when teams are close together and not much at all when teams are far apart). To me (and anyone with a brain), playing a stronger (or harder) schedule should mean playing games with a higher chance of losing. I calculate a team's SOS by estimating what the expected winning percentage would be for a hypothetical 45th-ranked team given a team's schedule, accounting for strength of schedule and location of the game. A team playing a schedule where the 45th-ranked team would be expected to win 70% of the time would be deemed easier than a team playing a schedule where the 45th-ranked team would be expected to win 60% of the time. Simple enough.

So, let's look at those 4 teams again.

Overall Expected Win Pct and SOS
Arizona (RPI SOS 1)- JCI SOS .5151 (1st)
Tennessee (RPI SOS 2) - JCI SOS .6386 (37th)
Syracuse (RPI SOS 13) - JCI SOS .6547 (53rd)
Washington St (RPI SOS 58) - JCI SOS .6108 (17th)

Non-Conf Expected Win Pct and SOS
Arizona (RPI SOS 5)- JCI SOS .6065 (21st)
Tennessee (RPI SOS 7) - JCI SOS .6370 (39th)
Syracuse (RPI SOS 19) - JCI SOS .7805 (203rd)
Washington St (RPI SOS 227) - JCI SOS .8025 (241st)

A few things jump out of the page. First, how can Washington State's RPI SOS be so much worse than Tennessee's despite playing a tougher schedule. And second, how can Syracuse's NC RPI SOS look so much better than Washington State when it's really not that much better. Let's look at each in more detail.

Tennessee is benefitting from the bottom feeder effect. They've only played one pure cupcake (and RPI killer), Prairie View A&M at home. Both Tennessee and Washington State played North Carolina A&T at home this season. The difference is Tennessee only played one non-conference game easier than that, while Washington State played five.

The cupcakes weigh down Washington State's overall numbers. If you split each team's schedule between the toughest 15 games and the rest, you'll see a distinct difference between the Vols and the Cougars.

Expected Win%, Toughest 15 games
Tennessee 7.10 - 7.90, 47.35%
Washington St 5.30 - 9.70, 35.30%

Expected Win%, Remaining Games
Tennessee 8.86 - 1.14, 88.64%%
Washington St 10.59 - 0.41, 96.23%

Wahington State is risking 1.8 more wins on the tough part of the schedule in exchange for risking 0.8 less wins on the easy portion of their schedule, and this difference causes Tennessee to have a RPI SOS of 2 and Washington State to have a RPI SOS of 58.

With respect to the non-conference portion of the schedule, Washington State is hurt relative to Syracuse by both the bottom-feeder effect and the lack of recognition for the location of the game. Washington State's three toughest non-conference games were @Gonzaga, @Baylor and @Boise State. Syracuse only played one road non-conference game and its three toughest were neutral games vs Ohio State, vs Washington and on the road against Virginia. If you stack the two teams side-by-side, Washington State played three of the four non-conference games between the two teams, but this is more than offset by the fact they played eight of the nine easiest non-conference games. That's how you explain the difference between NC RPI SOS of 19 vs 227.

And with respect to Arizona, I can't really disagree with their #1 overall SOS. They have played the toughest schedule to date. But like I showed in part three of my presentation, having that SOS creates a win-win situation for them. Do well and you are rewarded with a high seed. Don't do quite as well (like this year at 15-10) and you're still likely to make the tournament. I think it really says something when Arizona risked about 2 1/2 more games with its tougher schedule relative to Washington State, yet Washington State needs to go 21-5 to have the essentially the same RPI as Arizona at 15-10.

So maybe the bigger question is this...what is Washington State thinking? They did the same thing last year, playing a cupcake non-conference that dragged their RPI SOS down to 88 (versus my measure of 43) and overall RPI to 25 (versus 13). Why go into Selection Sunday with deflated numbers when it would be just as easy to go into Selection Sunday with more accurate (or even inflated numbers). It didn't hurt them last year as they did get a 3 seed, but it's not a good long-run strategy. Why haven't more programs caught on to this? Tournament appearances, Final Fours and National Championships are obviously goals for each of these schools, so why not best position yourself to accomplish these goals? Improved RPI and SOS numbers can never hurt you, they can only help you. If the difference between making the tourney and not making the tourney or landing a 6 seed instead of a 7 seed means playing the 200th-ranked RPI team at home instead of the 300th-ranked RPI team, I think you have to risk that one-thousandth of a win.