Analysis: Has the offense or defense improved?

A question that has come up a lot recently on HoyaTalk is whether the team defense has improved over the course of the season.

We can use the Game Plan data to take a look (click on image to enlarge):

This plot shows the raw offensive and defensive efficiencies for each game played so far. From the plot, it looks like there has been an improvement in offense over the course of the season (multiply the slope by 30 days (~month) and you get ~5 ticks more efficient / month). Defensive efficiency, unfortunately, is also rising, albeit at a slower rate (~2.3 ticks / month), which would mean the defense is actually getting worse. Since G'town averages about 60 possessions / game, a "tick" ( = 1 point / 100 possessions) represents about 0.6 points, so Pomeroy suggests that the offense is improving about 3 pts. month, but the defense is giving back about 1.4 pts month.

However, these numbers don't take into account the quality of opponent. We can do that by multiplying G'town's Off. Eff. by the opponents Adj. Def. Eff., then dividing by the season average (101.7 on Feb. 19th), to produce a net offensive efficiency. This represents how well the Hoyas' offense performed, accounting for the average defensive effort of the opponent. The same can be done for G'town's defensive efficiency (note here that I've flipped the sign for Net Def. Eff., so that a positive number mean G'town played better defense than average against their opponent). Here we go:

Well, things look a bit different now. The offense, accounting for the better defensive teams played lately, is actually getting much better as the season goes on (~5.3 pts / month), while the defense is actually stagnant (0.1 pts / month). What else is obvious is that there are 3 games since Jan 1st where the defense has been lousy [@ Pitt - L (1/13), Cinci - W (1/27) and @ L'ville - W (2/7)], but for all other games since early December, the Hoyas have defended about as well as or better than average.

Now there's a lot more you can do with these stats. For instance, do Net Off. Eff. and Net Def. Eff. track? This could indicate either playing to the level of your opponent, or my pet theory, the influence of the refs on the game:

And there does seem to be a trend, although not as significant as expected. When the Hoyas play lousy on defense, they play better on offense, and vice versa. Hard to say if this is just the result of how tightly the game is called, or another factor. I'll need to look at this more closely down the road.

We can also break out offensive and defensive efficiencies into their constituents, and look for trends. Here, I'm using the same technique, comparing how well G'town performed in each of the four factors against the average performance of their opponent.

Offensive Efficiency Components

Defensive Efficiency Components

For these charts, the correlations on the left are eFG%, TO% on the top, OR%, FTR on the bottom. Now admittedly, these are a bit of a mess (as always, click on image to enlarge), but from the top chart, you can see that 3 of the four factors are improving; only free throw rate (the number of FTs per possession) has remained flat.

Defense is much more interesting. The most important factors, eFG% and turnovers created, have remained flat, but Georgetown seems to be giving up more offensive rebounds while fouling less. Not sure what to make of it; perhaps the team is playing more zone defense, which is reducing fouls at the expense of rebound position, but that is just conjecture.