Pages

Most Deserving (The Case for George Washington)

3/07/2006

I've been catching a lot of flack for having George Washington on the '1' seed line in my bracket projections (starting last week and again in my most recent JCI rankings). However, I just wanted to clear the air and let everyone know that I don't honestly believe that they will get a #1 seed, for one or all of the following reasons:

  1. Their RPI numbers are artificially low, which can only hurt them when the Committee looks at their profile
  2. With their best player out, I think it's highly likely that they will lose in the Atlantic 10 tournament, dropping them from one-seed consideration
  3. Even if GW wins out, some other team from a major conference (Illinois, Ohio State, Texas, Memphis) will compile enough quality wins in their respective conference tournament to overtake George Washington

But if you believe that the Selection Committee is truly selecting teams based on their 'body-of-work', i.e. their complete resume of wins and losses over the course of the season, then George Washington definitely deserves consideration. Their JCI ranking, which is designed purely with that body-of-work objective in mind, of #4 bears that out.

So why the uproar? How can the #4 JCI team be a #4 seed in the minds of many bracketologists and hoops experts? Do they simply believe that George Washington isn't that good? It's true that there is a possibility that it's purely a fluke that GW is 26-1 -- that they're really a much lower ranked team that happened to get lucky time and time again. Given their relatively easy schedule (JCI SOS of 172, it's conceivable that a large number of teams could go 26-1 under GW's schedule. It is true that there is a wider margin of error around GW's ranking versus a team like Villanova given the lower number of quality opponents, making it more difficult to assess what GW's 'true' ranking really is. These statements are all true. However, it is equally likely that the complete opposite is true, that GW's is actually better that their record indicates. The only fair thing to do is to take the best estimate, which is what the JCI is doing.

I view the regular season as 30-35 opportunities for each team to show how good or bad they really are. And they do that by winning or losing. Quality teams should win their fair share of games. Simple as that. Given the same schedule, a 28-0 team should be considered better than a 27-1 team, which should be considered better than a 26-2 team, etc. Under a true body-of-work analysis, this is the crux of what the Committee is trying to do -- deciding if Team A at 21-10 is more worthy than Team B at 25-8 given the fact that they've played vastly different schedules.

The problem is that wins and losses are sticky things. You could have two evenly matched teams that might expected to go 5-5 if they played 10 times on a neutral court. However, they may only play once and it's a win for one team and a loss for the other. It'd be great if each team played a 3-game series like in baseball or played a round-robin against each of the other 333 Div I teams (both would eliminate the need for rankings or a committee altogether, win-loss record would be enough), but that's not going to happen anytime soon. Still, 30-35 games is enough to make a solid judgement for each team, and given the basic objective of a game is to win, you have to give credit to the team that wins and penalize the team that loses (what kind of can or worms would we be opening if we started discounting wins or adding losses to a team just because you didn't think they were that good...how would that be considered fair?)

So is there any way we can put some structure around the term 'deserving' and see if that sheds any light on the situation? How do we define whether a team is 'more deserving' of a 1 seed or an at-large bid than another? I'd submit this simple definition. Consider the comparison between UConn (27-2) versus George Washington (26-1). UConn would be considered more deserving if a xth-ranked team (take the average of the two teams in question) would be expected to have a more difficult time winning 27 games given UConn's schedule versus winning 26 games given George Washington's schedule. In other words, UConn's 27-2 record would be considered more impressive than GW's 26-1. Isn't this basically what the Committee is trying to decide?

With the help of the JCI, we can easily calculate and estimate these probabilities for any two teams. I ran 50,000 simulated seasons for each pair, and in the example above the xth-ranked team (again, the midpoint between the two teams in question would win 26 games under GW's schedule 87.2% of the time, versus winning 27 games under UConn's schedule 44.9% of the time. No surprise, but clearly UConn would be considered more deserving (I'll express this advantage as a difference in percentages: +42.2%.

So how does GW stack up against other top 10 teams?

George Washington (26-1) versus:
Connecticut....... -42.2%
Villanova......... -27.1%
Duke.............. -21.1%
George Washington. 0.0%
Memphis........... 9.8%
Gonzaga........... 14.3%
Illinois.......... 15.7%
Ohio State........ 16.5%
Texas............. 19.9%
North Carolina.... 35.7%

The numbers show that George Washington 'deserves' to be compared favorably to teams such as Memphis, Gonzaga, Illinois, Ohio State, Texas and North Carolina. In fact, even if you assumed that GW had a 25-2 record with the same schedule, they would still be considered more deserving (+7.2%) than North Carolina for 10th on the S-Curve. Saying that they should be a 4 or 5 seed would be like taking 3-4 of their actual wins and counting them as losses.

A couple of other interesting teams to look at:

Gonzaga (27-3) versus:
Duke.............. -42.6%
George Washington. -14.0%
Memphis........... -6.7%
Illinois.......... -2.7%
Ohio State........ -0.2%
Gonzaga........... 0.0%
Texas............. 3.9%
North Carolina.... 23.2%
Tennessee......... 28.5%
LSU............... 36.3%

This is where the margin of error on the easier schedule does come into play. Illinois and Ohio State are both considered 'more deserving' than Gonzaga despite having lower JCI's. I ran 100,000 obs for each just to make sure. The other interesting note about Gonzaga is they should be considered very similar to Memphis -- both 27-3 w/ similar SOS's (Memphis 79th, Gonz 97th), yet you never hear any consideration of Gonzaga being a 1 seed unless your name is Seth Davis. Certainly they aren't that far apart. (by the way, Gonzaga at 26-4 would still be considered better than Tennesse and LSU).

Illinois (25-5) versus:
Duke.............. -46.1%
George Washington. -15.8%
Memphis........... -5.0%
Illinois.......... 0.0%
Ohio State........ 2.0%
Gonzaga........... 2.3%
Texas............. 7.6%
North Carolina.... 31.5%
Tennessee......... 37.4%
LSU............... 46.2%

Ohio State has the Big Ten title, so that's probably enough to carry them above the Illini on the S-Curve, but you can begin to see the impact of Ohio State's easier conference schedule. Certainly, it's not out of the question that Illinois could finish higher than Ohio State depending on what happens in the BTT. You could even make a case for it today.

Here's one team on the bubble. I won't tell you immediately who it is

UAB............... -14.0%
Southern Illinois. -12.1%
UNC Wilmington.... -11.3%
Louisville........ -10.8%
Alabama........... -9.0%
George Mason...... -8.8%
Creighton......... -4.4%
Colorado.......... -3.3%
Bradley........... -3.1%
Air Force......... 0.6%
Vanderbilt........ 5.8%
Notre Dame........ 10.0%
Clemson........... 14.7%
California........ 18.4%
Miami-Florida..... 25.6%
Saint Joseph's.... 26.3%
Virginia.......... 27.9%
Iona.............. 28.0%
Northwestern...... 30.6%
Rutgers........... 31.1%

The team is Arizona. Now, I love to pick on the Wildcats simply because it's remarkable to see what a grossly inflated RPI and RPI SOS (thanks to some shrewd scheduling) can do for a team's tournament resume. You don't see anyone talking about Louisville being a lock much less a bubble team, and Louisville's season stacks up favorably to Arizona's.

One more...Joe Lunardi has Utah State and BYU listed as bubble teams (they've been on his Next 4 Out List for awhile) and I can't quite figure this out. Utah St (20-7) is 73rd in the JCI and BYU (19-7) is 74th. Hardly bubble teams. Let's look at how Utah State stacks up against the following teams.

Utah State (20-7) versus:
Miami-Florida.... -18.5%
Saint Joseph's... -17.2%
Iona............. -15.8%
Virginia......... -15.8%
Northwestern..... -13.2%
Rutgers.......... -12.6%
Minnesota........ -7.8%
Western Kentucky. -6.8%
DePaul........... -5.6%
Penn State....... -5.6%
Akron............ -4.4%
Old Dominion..... -3.2%
Kent St.......... -3.2%
Northwestern St.. -3.2%
South Carolina... -2.0%
Utah St.......... -0.2%
BYU.............. 1.9%
Nebraska......... 3.3%
Xavier........... 3.4%
Temple........... 3.6%

I think it's safe to say that if you stack up unfavorably to Penn State, Northwestern and DePaul, then you're probably not a bubble team. C'mon Lunardi, what are you thinking?

0 comments:

 
Wegoblogger #31 © 2011 | Designed by Bingo Cash, in collaboration with Modern Warfare 3, VPS Hosting and Compare Web Hosting