Monday, May 4, 2009

Experience and Depth

One of the more commonly blamed culprits for this year's disappointment in Hoyaland was experience, and, as an extension, depth of the experience.

The Hoyas had only an average of 1.11 years of experience per player on the floor this year, according to Ken Pomeroy, which ranked 317th out of 344 teams in DI.

There were less or similarly experienced teams who were more successful than the Hoyas. Kansas (333), Ohio State (327), West Virginia (291), and Butler (341) all had very successful seasons despite being young.

Georgetown was also not the only disappointment. Mississippi (344), Florida (312), and Oregon (330) all had very talented teams with disappointing seasons. And Arkansas (340) had a very similar collapse to the Hoyas.

This sort of anecdotal comparison is not yielding anything definitive. It certainly does not disagree with the idea that young teams are inconsistent; that they can be affected negatively by lack of experience. But it does not definitively say that, either.

I also ran some correlations on a team's experience versus different results, such as the overall Pomeroy rankings, offensive and defensive efficiency, Pomeroy's luck factor, etc.

One major issue with running this kind of analysis is that in college basketball, younger players are not just younger, but they are more physically talented. The NBA draft and other professional opportunities pull the players with the most talent before they can become experienced.

The result is a biasing of the data. First, across teams, the most experienced tend to be mid and low majors, which have significantly less talent than high majors. And secondly, even within high majors, teams like Kansas and North Carolina have more talent and less experience, year in and year out, than many of their conference counterparts.

In an attempt to balance for talent, I've done two things. First, I limited my subset to just major conference teams. This should partially eliminate the first issue. Second, I looked at the year over year change for the same team (for all teams), assuming that generally, teams recruit the same level of talent year in and year out. This really limits the amount of data to look at (Luck was only calculated from 2007-2009, so it is two year changes) but it does give a different perspective.

The result certainly does not imply a strong correlation.

Using major conference teams only for seasons ending in 2007-2009, the correlation between experience and Pomeroy Pythagorean winning percentage is .21 and the correlation between year over year change in experience and year over year change in Pythagoran winning percentage is .31. These are not incredibly low correlations in the real world, but they are hardly fantastic indicators of a driving force.

Perhaps the most interesting thing is that Pythagoran Winning % has the highest correlation of the results I tested. Offensive Efficiency correlated at about .20 when I limited to major conference teams; when I did year over year it was below .1. Defense was much stronger in year over year (~.2 again), but was not as strong in the major conference grouping.

Perhaps more surprisingly, in neither data set was there a correlation above .1 for either the Pomeroy "Luck" factor or the Consistency factor.

It does not seem that younger teams are actually more inconsistent or more likely to lose close games.

This goes against much of conventional wisdom. I'd like to do an expanded study at some time but for the moment, what does this mean for the Hoyas?

It seems to me it does not bode that well. I personally was chalking up much of the Hoyas lack of late game execution last season (as well as a general lack of execution) to a lack of experience. I assumed that this execution would improve with experience. That simply may not be true.

On the other hand, the above does not mean it is not true. But we can say that youth doesn't seem to be a uniform cause of "bad luck" (losing close games) or consistency.