Defense = (blk+stl)/fl?

The Search for a Defensive Efficiency Stat
One of the nifty benefits of Dean Oliver's Individual player offensive rating (ORtg) is how the number is useful for identifying how well (efficiently) a player converts possessions into points. As most come to understand however, that stat needs a context -- what role (major contributor? significant contributor? role player? limited role?) does the player actually serve in the offense. That question needs one (or sometimes two) more numbers -- the percentage of possessions the player gets when he is in the game, and maybe also how often the player gets a shot (FGA). Want to compare the conversion proficiency of a player who takes most of his shots inside the arc with one who takes a large portion of his shot outside the arc? Use the eFG% ((FGM + 0.5 * 3FGM)/FGA). Want to know how efficiently a player registers points when he takes a field goal attempt? Use the PPWS or True Shot (Pts/(FGA + 0.475 * FTA)...for the PPWS, which can then be divided by 2 to get the TS). And know that the offensive efficiencies are more easily identified and measured. Dean Oliver, author of Basketball on Paper, devoted a specific chapter ("Individual Defensive Ratings") in his book, to discuss the difficulties encountered when trying to identify (and collect/compile) stats that would be useful to measure defensive efficiency.

Magic Bullet or Magic Marker?
A stats geek (BadgerCane) over at APBRmetrics, proposed a statistic that divided the sum of a player's blocks and steals by the number of fouls he committed. I ran the numbers for Big East players for the last 2 seasons, looking only at players who logged 450 or more minutes (about 37.5% of the playing time at their position) in each season (2007-08 & 2008-09). The Top 30 players over the two season time frame...

		2 Yr.
Player	College	BkSt/Fl
Hasheem Thabeet	UConn	1.875
Terrence Williams	LVL	1.439
Earl Clark	LVL	1.368
Rick Jackson	Cuse	1.275
Jonny Flynn	Cuse	1.123
Sam Young	Pitt	1.113
Dominic James	MU	1.068
Hamady N'Diaye	RU	1.033
Jerome Dyson	UConn	0.965
Stanley Robinson	UConn	0.955
Will Walker	DPU	0.948
Geoff McDermott	PC	0.940
Paul Harris	Cuse	0.938
DeJuan Blair	Pitt	0.936
Jerel McNeal	MU	0.934
Jeff Xavier	PC	0.921
Jeremy Hazell	SHU	0.899
Weyinmi Efejuku	PC	0.892
Jeff Adrien	UConn	0.868
Alex Ruoff	WVU	0.860
Jessie Sapp	GTU	0.860
D.J. Kennedy	SJU	0.856
Corey Chandler	RU	0.852
Eugene Harvey	SHU	0.843
Paris Horne	SJU	0.833
Dante Cunningham	Nova	0.814
Dwayne Anderson	Nova	0.811
Ryan Ayers	NDU	0.797
Arinze Onuaku	Cuse	0.778
Jerry Smith	LVL	0.755
Craig Austrie	UConn	0.754

All the Usual Suspects...and a Few Surprises
>1.0 is very good. If the Big East's Defensive Player of the Year (2 consecutive seasons) -- Hasheem Thabeet -- does not surprise, the fact that he registered nearly 2 blocks/steals per foul gives pause to think. Using the stat to measure magnitude (is Thabeet twice as "defensive" as Austrie?) is hardly appropriate, but the list is certainly studded with names known throughout the conference for their defensive abilities and athleticism -- James & McNeal from Marquette, Williams & Clark from Louisville, Pitts' Sam Young and DeJuan Blair (both gone to the league this post season), Flynn, Jackson, Harris and Onuaku from Syracuse, and virtually the entire Connecticut starting line up (Thabeet, Adrien, Robinson and Dyson...only Price, who was ranked #35, is missing -- small wonder the Huskies, along with the Cardinals were far and away the best defensive unit in conference games last season). Villanova's Dante Cunningham and Dwayne Anderson, two well regarded defenders on the Wildcat squad for the past two seasons, are also ranked in the top 30.

On the other hand when Jeremy Hazell and Eugene Harvey from Seton Hall are mentioned, it is generally not for their defensive exploits. Hazell is regarded as a prolific scorer. Weyinmi Efejuku, Will Walker, Jeremy Hazell all combined good (to excellent) steal rates with a lower than average propensity to commit fouls, attributes that suggest good defensive skills.

A Few Other Wildcats
A few of the other squad members who logged the requisite amount of time...

		2 Yr.
Player	Rank	BkSt/Fl
Scottie Reynolds	38	0.706
Reggie Redding	47	0.616
Corey Fisher	50	0.552
Corey Stokes	64	0.441
Shane Clark	66	0.390
Antonio Pena	71	0.341

For Reynolds and Fisher, whose steal rates were comparable to Will Walker's, Geoff McDermott's and Jeremy Hazell, the key was fouls. Both players were whistled often enough to diminish their ratings. The same holds for Pena and blocks -- the Villanovan had a block rate comparable to Geoff McDermott (his ranking was a combination of blocks and steals...) and DaJuan Blair, and better than Jeff Adrien. But the Wildcat also fouled at a higher rate than those three players.

Zebras and Cats: Getting Down to Cases

A Gary Parrish column, "Myth Busters", from a few weeks ago put me on to a feature from Robbie Allen's StatSheet.com website. I linked to Robbie's website during the season and have used it for a few of Villanova's stats over time, but I did not realize (until Parrish's column...) that StatSheet.com also collected stats on referees. I have been tracking the fouls, free throws and technicals for villanova's games the past few seasons. I found I could compare referees (or more specifically teams of referees) to each other in how they handled Villanova games. I realized I could now step back to see how these referees handled other games and teams they worked. Was (for example) Tony Greene "harder" or "easier" on Villanova than he was on other teams? The table below provides a few answers and points of comparison.

	All Games			Nova Games
Name	Gm	FPG	STD	Gm	FPG	STD
Hillary, Ted	89	33.8	7.1	1	51.0	NA
Rutecki, Wally	50	37.9	7.1	1	50.0	NA
Laubenstein, Bill	29	37.0	7.0	1	50.0	NA
Perone, Ray	70	37.4	6.7	1	47.0	NA
Hess, Karl	89	36.2	7.1	3	47.0	4.6
Clougherty, Tim	54	34.5	6.2	1	47.0	NA
Donato, Bob	60	37.5	7.1	4	45.0	5.3
Luckie, Jamie	93	38.3	7.1	4	44.0	3.9
Higgins, Tim	69	36.3	7.2	3	43.7	4.2
Kersey, Bryan	102	37.9	7.5	3	43.3	8.1
Stuart, Mike	77	37.1	6.7	2	42.5	4.9
Stephens, Michael	61	38.4	6.5	4	42.3	5.6
Cahill, John	90	35.9	6.1	7	42.0	5.4
Clinton, Bernard	60	40.2	7.5	2	41.5	10.6
Walton, Earl	67	39.1	7.2	2	41.0	5.7
Jones, Les	79	36.7	7.5	1	41.0	NA
Higgins, John	97	35.5	7.2	1	41.0	NA
Collins, J.D.	103	36.8	7.4	2	40.5	2.1
Driscoll, Pat	62	36.8	6.8	2	40.5	3.5
Haney, Jim	68	36.5	7.0	2	40.5	0.7
Greene, Tony	82	35.8	5.7	3	40.3	0.6
O'Connell, Brian	87	35.8	6.3	5	40.2	7.3
Shaw, Curtis	102	39.3	7.3	2	40.0	2.1
Hightower, Ed	80	37.1	8.0	2	40.0	12.7
Prager, Gary	57	36.3	5.7	1	40.0	NA
Burr, Jim	85	35.5	7.7	4	39.5	2.4
Janssen, Paul	56	37.4	7.7	1	39.0	NA
Greenwood, Reggie	65	36.8	6.2	3	38.0	3.0
Shows, Doug	72	36.4	8.1	2	38.0	4.2
Roberts, Mike	61	35.6	6.1	1	38.0	NA
Corbett, Ed	87	34.4	6.5	7	38.0	2.6
Lindsay, Joe	65	35.1	5.9	2	35.5	3.5
Kitts, Mike	87	35.0	7.0	4	35.5	3.7
Clark, Jeff	62	36.8	6.9	3	34.0	3.6

Those looking for stats on Tom O'Neill's Villanova game will be disappointed. The referees listed are those who worked Big East regular season contests -- O'Neill was on the zebra crew that worked the Cats - Golden Knights game in the 1^st round of the Old Spice Classic; O'Neill did not work Big East regular season games. My Villanova game stats however, do include those games against OOC opponents, however, so I am comparing the Villanova games against all games the individual referee worked. In general, the zebra's FPG was higher for the Big East games his crews worked than for all games. I highlighted (in yellow) those referees whose average FPG for Villanova games was outside of the range of the standard deviation for all of their games. In most cases the referee worked a single game, certainly reasonable. Of the three others, two are most likely recognizable to many Wildcat fans, Tim Higgins (recognizable to many basketball fans throughout the country) and Bob Donato, the referee who called Corey Stokes for a blocking foul on Jon Wallace near the end of regulation play in the Villanova-Georgetown game. Yeah, that Bob Donato. The only referee to have a lower average FPG for his Villanova games than he does for all games, Jeff Clark, is highlighted in green. The only referee whose average for Villanova games was lower than his average for Big East games, Doug Shows, is highlighted in brown. Clark worked 3 Villanova games, posting a 3-0 record for those games. Shows worked 2 Villanova games, posting a 1-1 record for those games. Other observations...
1. Ed Corbett and John Cahill worked 7 Villanova games each. That's more than any of the other referees. Villanova games accounted for 7.8% of Cahill's workload, and a straight 8.0% of Corbett's. Villanova's record in Cahill-refereed games was 1-6 last season. For Corbett, the record was 5-2.
2. The FPG for all games and for Villanova games was very close when Mike Kitts worked as part of the crew. At 7.0, Kitt's standard deviation was fairly close to the average for all 34 referees listed (6.9), but at 3.7, his crews showed more consistency when working Villanova games. The Wildcats were 4-0 last season when Kitts was in the crew.
3. Brian O'Connell was the third most prolific Villanova referee, working 5 games (5.75% of his overall workload). Averaging 40.2 FPG, O'Connell seemed to have been, relative to the other 33 referees, more in the "let 'em play" faction. But with a standard deviation of 7.3, O'Connell was pretty inconsistent.

Zebras and Cats: 2007-08 Edition

For a few weeks in February there was a strong chance that referees Jim Burr and Bob Donato, arms outstretched to signal foul and clock stoppage in the (literally) very last seconds of one possession games, would become the icons for Villanova's season. Count on a Sweet Sixteen run to soften the memory, but limping out of the MCI Center saddled with a 14-9 record, 2 of those losses coming complements of quick whistles, it seemed that garnering a fourth consecutive dance bid was slipping just beyond reach. A zebra conspiracy against all things Wildcat? Self-involved fans fixing on excuses for a young and over-hyped team's inability to perform to expectations? Fans who sensed "things were different..." with the referees were not hallucinating fanatics (those are the other teams' supporters...). But there is no statistical evidence (that I have found anyway) that suggests Zebras hate Wildcats. They just called more fouls on them last season. As the chart below suggests, the 'Cats reversed a 2 year trend that had opponents fouling at higher (per game) rates than Villanova. While the disparity was nothing on the scale of the 2004-05 season (also ironically a season that ended in a Sweet Sixteen game...), the balance tipped slightly against the 'Cats.

The very slight average per game advantage to opponents is taken on all games this past season. If the average was less than a single foul per game, the standard deviation suggested the number of fouls per game varied a bit more than expected. For opponents, the standard deviation was 3.39 on an average of 20.1 fouls, while fouls for the 'Cats, the standard deviation was 4.63 on 20.8 fouls, a bit higher than expected, and a little surprising when compared to opponents. Given changes in the starting lineup and rotation throughout the season, combined with the lack of collegiate experience the wide variation in fouls calling is understandable. Ken Pomeroy, per his Villanova Scouting Report calculated Villanova's "experience" at 1.0 years and ranked the Wildcats 324/341, a good indication of the team's relative youth (as a yardstick, consider that Pomeroy calculated the 2007 Connecticut Huskies at 0.5 years and ranked them 335/336).

Breaking the games out by location (home, away, neutral -- see graph below) reveals more detail, and some variations that I did not expect. "Home Cooking" was a myth last season. Villanova held a paltry 0.5 average foul advantage over opponents over the 15 game home slate. This is especially interesting because a team with a home court record like Villanova's last season, 12-3, ought to expect their opponent's foul total to be padded by a few extra change of possession fouls at the end of the game. The average margin of victory, 13.2, may have been large enough to discourage the practice. Despite the record and the modest advantage in average fouls the standard deviation revealed a similar gap in the range of the number of fouls called. Surprisingly, it came to nearly 2.5 fouls. The 'Cats actually did better on neutral courts, holding an advantage of nearly 1 foul (0.9) average over their opponents over the course of 9 games. The gap for the 11 away games explains the fouling disadvantage for Villanova. The Wildcats averaged nearly 3.5 more fouls per game than their opponents. Home Cooking in reverse (Road Curse?) seems appropriate. The Wildcats had a losing road record, 4-7, which may explain some of that difference, if one assumes that the team that is behind fouls 4 - 6 times in the closing minutes as a strategy to force possession changes and give themselves more scoring opportunities. The gap in standard deviations actually favored Villanova by a very modest 0.5 fouls.

Given the relative inexperience of the squad the struggles on the road could have been anticipated. The large inconsistency from game-to-game, even at home is puzzling though. Any comfort to be drawn from these numbers will have to come from the differential in the neutral court games. Villanova was able to hold their average below their opponents, despite the unfamiliar setting. That the neutral court games were played near the beginning of the season (Old Spice Classic) and at the very end of the season (GET and NCAA) should be a good sign for the squad next season.

Methods & terms...
I counted a game as "home" if it was played at the Pavilion or the Wachovia Center, as "away" if it was played on a home court of the opponent and as "neutral" if the game was played at a location not the home court of either team. This season neutral games were played at the Milk House for the Old Spice Classic (though one could argue that the proximity of UCF made the Milk House a virtual home court for the Golden Knights...), Madison Square Garden for the Big East Tournament (though I counted it as "away" for the St. John's game) and the arenas in Tampa and Detroit where the 'Cats played 3 NCAA games.