Wednesday, August 19, 2009

Experience and Close Games

Definitions and Methods
2009 was not notable for close games, at least not until the post season. A "close game" is one decided by 5 points or less. The 5 point margin suggests that the decision could have turned on 2 or fewer possessions -- a game that was probably settled in the last minute or so. Who would be more likely to play close games? A team with a bad record (like Notre Dame back in 2007)? I sorted the conference by percentage of the team's schedule that contained close games...

Record% ofExp.
TeamWLPctSchYrs
Seton Hall740.6360.3441.65
Rutgers560.4550.3441.39
Villanova720.7780.2372.04
Marquette440.5000.2292.45
Georgetown160.1430.2261.11
DePaul430.5710.2121.20
West Virginia340.4290.2001.26
South Florida150.1670.1941.58
Syracuse520.7140.1841.65
Providence330.5000.1822.53
Cincinnati410.8000.1561.29
Louisville410.8000.1351.72
Notre Dame310.7500.1112.48
St. John's301.0000.0880.86
Connecticut201.0000.0562.20
Pittsburgh110.5000.0561.94
Overall57430.5700.187

The table shows the team's record in close games, the percentage of the schedule that involved close games and lastly, Ken Pomeroy's Experience Level, a stat in which he combines minutes played with class. The Overall line sums wins, losses calculates the winning percentage and the close games as a percentage of all games played. [Note the winning percentage is > .500 because I have included all games -- OOC, conference & post season -- for each team]

Did "not very good teams" play more close games? Well, yes...and no. True, Seton Hall and Rutgers did have the most close games, about 1/3 of their schedule was decided by 5 points or less, but Villanova and Marquette, two teams that finished near the top of the conference were just behind. That set of two paired teams provide an interesting snapshot on their seasons, relative to each other. In close games Seton Hall pulled out over 60% of them, while Rutgers could not break even, a capsule statement of their respective seasons and records. The same holds for Villanova and Marquette. Many had those two teams in a virtual tie in the Big East standings going into the season, but the Wildcats held a very healthy margin in the close games. The Golden Eagles stood at 50% (four of those games occurred after Dom James ended his season with an injury; Marquette went 1-3 in those games...another effect of losing the senior guard?). While Connecticut and Pittsburgh, two 1st quartile teams had the fewest close games (Pittsburgh's in particular, were badly timed), but St. John's and Notre Dame, two teams that finished in the bottom half of the conference played only a single close game more than UConn and Pitt.

The two "losingest" teams, Georgetown and South Florida really caught my eye. Among teams that had 20% or more of their schedule decided by 5 points or less, those two had strikingly poor winning percentages. While the folks over at the Hoya Prospectus Blog have examined the Georgetown season in detail in an attempt to construct a narrative of the 2009 season, I have not seen a post that explains this abnormally low winning percentage, though a post on experience discussed overall performance (and concluded it was not a decisive factor). If Villanova is an outlier (on the high side), those two were clear outliers on the low side. Ironically one of Villanova's close losses and Georgetown's single close game win -- was the 2/28 game played at the Wachovia.

Learn Anything in the School of Hard Knocks?
Georgetown was one of the "younger" teams in the Big East. Was their bad run (as suggested by a few fans & examined in depth by statistical analysis) a product of their youth? The Hoya Prospectus Blog (see above) says "no", but my question is more specific -- was youth a factor when the game was close? Looking at all of the Big East teams (same set of games), for each close game, I compared the experience-level of the two squads to determine if the "more experienced" squad won. Going back to each team, I reclassified each close game according to whether the result was consistent with the experience level of the two teams -- ie "did the more experienced team win?". The survey yields a few interesting nuggets...

W/L vs ExpExp.
TeamAg.Dis.Pct.Yrs.
Syracuse701.0001.65
Connecticut201.0002.20
Pittsburgh201.0001.94
West Virginia610.8571.26
Cincinnati410.8001.29
Notre Dame310.7502.48
Rutgers830.7271.39
Seton Hall830.7271.65
Louisville320.6001.72
DePaul430.5711.20
Georgetown430.5711.11
Villanova540.5562.04
Marquette440.5002.45
Providence330.5002.53
South Florida240.3331.58
St. John's030.0000.86
Overall65350.650

I sorted the data by the "expectations percentage" and included Pomeroy's experience level as a point of comparison. Hmmm, these results look pretty interesting. True Connecticut and Pittsburgh offer a sample too small to draw conclusions, but Syracuse and West Virginia look pretty conclusive. Seton Hall and Rutgers show consistent results (even though their won/loss records in close games were very different). Even down-on-their-luck DePaul and contrarian Georgetown show reults that agreed more than disagreed with the experience expectation. Looking at the bottom of the table, St. John's and South Florida appear to be counter intuitive (but consider that two of St. John's counter-expectation wins came versus Georgetown...that team again).

Marquette and Providence came at .500, both for winning percentage in close games (4-4, 3-3 respectively) and for "experience result consistency". They caught my eye because they are two of the more experienced teams in the conference -- Marquette was ranked #1 in the Big East according to Pomeroy, while Providence was #3, behind Notre Dame. Common sense (and this data) suggest they underperformed. Both also had 1st year coaches, relatively inexperienced 1st year coaches. Buzz Williams and Keno Davis each logged a single season as head coach before coming into the Big East.

No comments: