For years and years, I have yelled at my television set, the walls of my apartment, my coworkers and random people on the street about the need for the Iowa men’s basketball team to improve defensively.
A frequent concern is the idea that they practice too many defenses and, as a result, don’t really master any of them. Another concern is that we have certain players who are intrinsically very weak defensively, due to athletic limitations. Some of these same players are very good on offense, so benching them trades defense for offense with little net change in the outcome of our games. Thus we are just going to be stuck with good offense and lousy defense until these players graduate.
I have attempted to empirically measure the effectiveness of our various defensive schemes and all of our regular rotation players in the Minnesota and Michigan games. I did this by watching the games, recording the type of defense that we were in on each possession as well as the number of points that the other team scored on each possession. I also recorded all of the lineup combinations that Fran employed during a game and measured their effectiveness in an identical manner. From this, I determined the number of possessions that each player was on the court and calculated how many points we allowed per possession (PPP) when that player was on the court vs. when he was off the court. So there’s no attempts to extrapolate box score numbers into player value. Guys like Bart Torvik (free) and Ken Pomeroy ($) already have great tools for that which I highly recommend looking at if you are into this kind of stuff. My measurements are purely “what happened when we did this?”. You may notice that my point totals don’t add up to the actual point totals in the Minnesota game. This is because I omitted the last minute or so of the Minnesota game as it was desperate scrambling/fouling and isn’t representative of our overall defensive style.
I will share the individual defensive performance data after I’ve had more time to accumulate data from additional games. It is already clear that the data will be highly variable from game to game. Case in point: Jordan Bohannon. Against Michigan, we gave up 0.65 PPP when Bohannon was in the game and 1.5 PPP when he wasn’t. That’s right! Our defense was 0.85 PPP better when Bohannon was on the floor against Michigan. That’s an enormous difference. According to Torvik’s adjusted efficiency, the best defensive team in the country is Virginia at 0.84 PPP (adjusted) while the worst is Savannah State at 1.19 PPP (adjusted), a difference of just 0.35 PPP. So maybe if Savannah State just had Bohannon, they’d be Virginia. Still, I think I should gather more data before proclaiming Bohannon the best defensive player in collegiate history. For example, against Minnesota, we gave up 1.32 PPP when Bohannon was in the game and 1.17 PPP when he was on the bench. Although individual defensive efficiencies aren’t going to be a perfect metric of defender quality for a host of reasons, I am optimistic that it will give us useful information. It is just going to take several games to stabilize.
I did notice some interesting things in the defensive scheme data, starting with the Minnesota game. Every defensive possession was categorized into one of three types: man, zone, and transition. Transition defines a broad class of possessions that share the general characteristic of not lasting long enough for me to determine what type of half-court defense we were going to run. They include fast breaks, rapidly developing secondary breaks (i.e. the defense is back but the shot goes up before much cutting takes place so you can’t determine the defensive scheme) as well as turnovers that occur when we press.
Table 1 lists the three defensive possession types (man, zone, transition) for each half of the Minnesota game as well as the number (and net outcome) of possessions where we pressed or didn’t press. A number of things stand out. First, our first half defense was absolutely putrid against Minnesota. We gave up 1.53 points per possession (PPP). We ran a bit more zone than man. Neither was effective although our zone was not as astronomically ineffective (1.5 points per possession) as our man to man defense (2 points per possession).
Let’s all take a moment and think about that 2.0 PPP number. Shaq going up against toddlers would dunk every time. That comes out to the exact same points per possession as what our man to man defense allowed in the first half against Minnesota. The only possession type where we had any type of success at all were transition possessions, where we held Minnesota to six points on seven possessions. We only attempted our press on four first half possessions giving up two points total.
|Minnesota First Half Overall||36||55||1.53|
|Minnesota First Half Transition||7||6||0.86|
|Minnesota First Half Man||11||22||2|
|Minnesota First Half Zone||18||27||1.5|
|Minnesota First Half Press||4||2||0.5|
|Minnesota First Half No Press||32||53||1.66|
|Minnesota Second Half Overall||31||29||0.94|
|Minnesota Second Half Transition||13||14||1.08|
|Minnesota Second Half Man||14||13||0.93|
|Minnesota Second Half Zone||4||2||0.5|
|Minnesota Second Half Press||15||11||0.73|
|Minnesota Second Half No Press||16||18||1.125|
The second half was a very different story. For starters, we employed our press on 15 of the 31 recorded second half possessions. The pressure was quite effective as Minnesota only scored 11 points on those 15 possessions (0.73 PPP). Many of those press possessions ended in a transition possession (either due to turnover or quick shot attempt) but when Minnesota did break the pressure or, more precisely, when Minnesota faced our half-court defense, we primarily played man to man (only four second half possessions were zone). Our man to man defense was far more effective in the second half allowing only 0.93 PPP. Our defense on possessions where we didn’t press was 1.125 PPP in the second half, so the press seemed quite effective even in the context of an overall improved defensive effort / colder Minnesota team. 1.125 PPP isn’t great but is still considerably better than our non-press first half defense, which allowed 1.656 PPP. The improvement in our defensive PPP in the second half probably had many causes: our effort level, our scheme, Minnesota’s players relaxing a bit, Minnesota’s shooting regressing to the mean, etc. Still, we clearly made some defensive adjustments (specifically increased use of full court press and increased use of an improved man to man defense) and those adjustments seemed to work.
The script was flipped in the Michigan game. I analyzed the Michigan game before I analyzed the Minnesota game and I didn’t rigorously track the number of possessions on which we pressed in the Michigan game so I didn’t include that. There were a couple interesting things to note from the Michigan game as well in terms of defensive scheme, shown in Table 2. First, over the course of the game, we employed about an equal amount of man and zone defense and both defenses were approximately equal in effectiveness (0.87 and 0.94 PPP, respectively). Second, we ran more man to man in the first half and more zone in the second half. Our man to man defense was significantly more effective in the first half (0.7 PPP vs. 1.2 PPP in the second half). The drop in man to man effectiveness probably precipitated Fran’s shift to zone defense. According to my charts, we ran zone on the last 13 possessions of the game, during which time we surrendered a total of 6 points. That’s a good way to close out a game.
|Michigan First Half Overall||36||29||0.81|
|Michigan First Half Transition||6||3||0.5|
|Michigan First Half Man||20||14||0.7|
|Michigan First Half Zone||10||12||1.2|
|Michigan Second Half Overall||34||30||0.88|
|Michigan Second Half Transition||2||0||0|
|Michigan Second Half Man||10||12||1.2|
|Michigan Second Half Zone||22||18||0.82|
|Michigan Full Game Overall||70||59||0.84|
|Michigan Transition Full Game||8||3||0.375|
|Michigan Man Full Game||30||26||0.87|
|Michigan Zone Full Game||32||30||0.94|
Collectively, this limited data set has a couple of salient, albeit extremely preliminary, points. The first is that Fran’s philosophy of switching defense seems to have merit. It allows Fran the flexibility to make adjustments and find a defensive scheme that works. Against Minnesota, Fran’s increased use of the press helped out dramatically. Against Michigan, our switch to zone in the last 7-8 minutes was extremely effective. Obviously, this doesn’t prove anything. We still might be better off using one defense and trying to master it. But it does show that these defensive switches can work.
So that’s all for now. I plan on analyzing our remaining games in the same manner and might go back into some of the older games as well. Hopefully, there will continue to be interesting things to be learned from this. In particular, I’m looking forward to seeing the individual defensive efficiencies once we get a few more games. These analyses are still in their early stages so if there are ways that I can do it better, please speak up!