If you’ve watched Toledo play (or even listened to a radio broadcast as I did for their game against Stony Brook earlier this season), you know that the Rockets play an extremely exciting brand of basketball. Their offense is awesome (10th in adjusted offensive efficiency according to Ken Pomeroy) and the defense isn’t particularly good (261st). Every game is a shootout. As I looked at those numbers I started wondering, “Which teams rely the most on either their offense or defense?” So I made up a stat to figure it out!

Just subtracting adjusted defensive efficiency rank from adjusted offensive efficiency rank wasn’t going to be enough to figure out what teams were the most reliant. It had the potential of “middle of the pack” teams with a big disparity screwing up the rankings. So instead I decided to use this formula to figure out the teams that rely the heaviest on their offense:

(Adjusted Offensive Efficiency Rank – Adjusted Defensive Efficiency Rank) / (Adjusted Offensive Efficiency Rank + Adjusted Defensive Efficiency Rank)

Turns out this did a pretty good job! A negative ranking in this statistic means that you’re heavily dependent on your offense (because your Adj. Off. Eff. Rank is greater than your Adj. Def. Eff. Rank). Here is the Top 10 (Offense Rank, Defense Rank):

- Gonzaga (2, 68)
- Duke (3, 90)
- Toledo (10, 261)
- Boston College (23, 298)
- Creighton (6, 74)
- Oregon (8, 86)
- Oklahoma (17, 145)
- St. Mary’s (11, 93)
- Kentucky (5, 40)
- Lafayette (45, 315)

This list certainly passes the smell test. Why just the other day people were writing about Kentucky’s defensive struggles. People have been worrying about the Duke defense since the season began or at least since the narrow home victory over Vermont. Get ready for some shootouts when the West Coast Conference’s two titans face off! That should be a ton of fun. I’d also highly recommend that you go check out any game involving any of the 10 teams on this list. They make basketball lots of fun by allowing both teams to light up the scoreboard. Another local favorite, Iona, isn’t much farther down the list at #19.

What about teams that have high positive rating, meaning a strong defense relative to their offense? Here’s the Top 10 from that list.

- Saint Louis (133, 3)
- Ohio State (35, 1)
- Clemson (128, 4)
- Villanova (38, 2)
- Virginia (86, 5)
- Kansas St. (224, 21)
- UNLV (254, 25)
- Western Illinois (340, 47)
- Indiana (93, 16)
- St. Francis Brooklyn (312, 59)

All you have to do is watch one of the Terriers’ games and you know why they’re here. When three-point shots aren’t going SFC struggles to score in any gym, but they managed to shut down Syracuse and Miami (FL) on the defensive end. The team has a ton of length and athleticism, it just hasn’t translated on the offensive side of the ball yet. You might want to bring a pillow to Clemson, Virginia or Western Illinois games. Not only are those teams on this list, they also play at extremely slow paces (347, 337 and 339 in adjusted tempo respectively). The other place you might need a pillow? Some Big East games, if only for protection. There are going to be some rock fights their this season. Beyond the Wildcats, four teams currently rank in the Top 35 nationally in this metric (Xavier – #12, Georgetown – #17 and Marquette – #31). It’s worth noting that the Big Ten has three (the two in the Top 10, plus Michigan St. at #14).

Because made up stats are fun. Here are the Top 5 in both categories since 2009:

**Offense Heavy:**

- 2012 Missouri
- 2013 Michigan
- 2010 Cornell
- 2011 Notre Dame
- 2011 Oakland

**Defense Heavy:**

- 2010 USC
- 2011 Florida St.
- 2012 Louisville
- 2013 Wisconsin
- 2011 St. Peter’s

TWCDefensive based teams are still at a disadvantage because of the new rule changes.

I would think that would be reflected in the foul shooting advantage which should be put in the offense heavy catagory.

Pingback: The Secondary Break: Friday’s Links | CollegeBasketballTalk

FloormasterThis is a mathematical/logical mistake. You should be using the actual adjusted ratings instead of the ranks (ratio should be cardinal not ordinal). SLU and Ohio State have good offenses as well as good defenses and while they are more reliant on their defense (ranked properly 25th and 33rd respectively) a team like Western Illinois is more defensively reliant as their offense is terrible and defense is well above average.

Email me if you want my calculations but you should redo this (hot hard)

John TemplonPost authorUsing the actual adjusted ratings instead of the rankings produces much poorer lists in terms of results. They’re much less intuitive and would require additional scaling. Using the rankings on both sides makes more sense editorially for sure. I’m not saying that Ohio State or Saint Louis’ offenses are “bad” per say, they just aren’t nearly as good as their defense compared to the rest of Division I basketball. Whereas a team like Louisville is strong on both sides of the ball. Also, using rankings makes the ratio intuitively scale from 1 to -1, which makes it easy to see where teams rank. If you wanted to adjust to a ratio of how a team is performing versus the average on both adjusted ratings that could make sense. The fact is though that OSU and SLU both rely heavily on their defense to win basketball games.

FloormasterYou missed the point. You can use a scale from 1 to -1 if you want the problem is you are using ranks, ORDINAL numbers in ratios. You are essentially saying by dividing a rank that the 1st team is 10 times better than the 10th team (that is incorrect).

Yes, Ohio State and SLU rely on their defense; I said that as well as properly calculated they would be 33rd and 25th respectively in this Defense to Offense ratio. The important point is that they are just not one and two which is simply an artifact of having a low rank/ordinal number in defense.

Actually your “rank ratio” would probably make Louisville appear more offensive minded as they have a 4 to 1 ratio. If you are saying that Louisville is balanced (they are) then your method inflates their offensive mindedness (properly calculated Louisville is around 90th in offensive mindedness).

John TemplonPost authorWhat’s your “proper calculation”? As you can see above, I’m not in any way trying to say Louisville’s offense is 4X better than it’s defense. The numbers don’t say that either. It’s exactly the reason I’m not doing a straight Offense Rank / Defense Rank ratio. Instead I’m doing (Offense Rank – Defense Rank) / (Offense Rank + Defense Rank). When I substitute the Ratings instead into that formula you don’t get good results at all. Can you please explain the formula you’re using where you claim Louisville is 90th? I’m certainly open to other opinions, I just want to understand your methodology.

FloormasterI apologize for the Louisville example as I forgot that you divided through by the sum of the offensive and defensive ranks (that causes other logical problems but I will let that pass). However, the rest of the point stands (however badly explained).

The calculation I did is a little complicated because the offense and defense are measured in reverse (with offense, higher is better and with defense, lower is better). So my calculation is the absolute number of the difference of their efficiency rating from the mean. You can calculate it yourself (I have a spreadsheet but no way to share it easily). You do not need a ratio as the numbers are on a consistent scale.

So, Pomeroy’s numbers change every day so let’s use todays numbers and look at the offensive “reliance” rank. Now your top ten are basically the ten you have up at the top (minor differences today; Duke changed places with Gonzaga, etc.). The way I did it they all show high offensive reliance ranks (in fact 5 of the top 10 are the same so no doubt this is why it passes the “eye test”). The biggest problem with your offensive reliance ratio rank is Kentucky for all the reasons I mention above are not top 10 in offensive reliance but 54th (they get in the top 10 simply because their offensive rank is low not because their offense to defense reliance is top 10).

Fundamentally, you cannot create RATIOS out of ORDINAL numbers. This is a really basic point that I cannot help you with any more (if I knew you better I could think of better examples). Read something like this: http://www.measuringusability.com/blog/interval-ordinal.php (related but not exactly my point). I guess I just found it weird you used ranks when the actual numbers are right there and make your points far more accurately.