Micro-Analysis #1: UCONN

If you are an avid reader of my blog, you know I have a tendency to look at the college basketball world from a top-down perspective (a.k.a. - the macro approach). You will also know that I keep saying I want to do more micro-analysis on the blog and then never do. Since I'm still tying together loose ends on other models, this seems like a better-now-than-never opportunity to do a micro-analysis article. Though it may seem far different than the wordy and explanation-of-the-explanation-of-the-explanation style of my other articles, I do want these articles to be quick reads and the numbers to speak for themselves. So, let's dive into it.

The Curious Case of Crashing UConn

Before Christmas and any of my QC analysis, CONN looked like the undisputed champ of college basketball. They were winning games by 20+ points, two of which were against current Top 15 teams, and they were beating teams by 20 points that other good teams were only winning by 10 or less points. Then, after their eight day Christmas break, things went south quickly, and the #1 AP-ranked team has now found itself hoping to stay in the Top 20 by next week's poll reveal. Let's start with possessions.

A quick word on notation, and then to the details. Green is CONN stats, Yellow is Opponent stats unless actually labelled Opp (Opp Pts), Blue is Possessions (a shared stat), and Mg is Margin (the first is point margin and the second is turnover margin, where positive is good and negative is bad).

1. Possessions fell 1.8 per game from 70.9 to 69.1. Possession affect margin of victory, assuming efficiency stays the same. Suppose you have two teams where one averages 1 point per possession and the other averages 0.95 points per possession. Then, a 60 possession game will result in a final score of 60-57, whereas a 100 possession game would result in a final score of 100-95.
2. In the first block of games, CONN only had one game below 1.07 points per possession, which was the first BEC game at BUT. In the second block of games, CONN only had one game above 1.07 points per possession, which was the loss at MARQ.
3. Turnovers haven't helped their cause. While TOs have slightly ticked up to 12.6 from 12.0, opponent's TOs have ticked down drastically from 15.5 to 11.6. Turnovers usually lead to higher efficiency transition buckets. Instead, fewer TOs result in less high-efficient shots for CONN and more actual shots for the opponent.
4. When you factor in the decline in the number of possessions, CONN's TO rate (not shown, but it is turnovers per possession, not per game) is 1% higher and their opponents is 5% lower.

Next, let's look at the shooting numbers.

Since this chart is a lot to absorb, let's pick up from where the possession analysis left off.

1. CONN's shot quantity hasn't changed too much, given the 1.8 decline in possessions per game.
2. CONN's shot quantity differential (their shot quantity compared to opponent shot quantity) has changed though, as opponent's are getting more shot attempts from fewer turnovers. CONN is shooting 1.5 less shots per game and their opponents are shooting 2.2 more shots per game.
3. Shooting percentages are going the wrong way on both sides of the ball for CONN. 2P% is down almost 10%, 3P% is down 5%, and opponents' 2P% is up 10%.
4. Since shot quantity is roughly neutral for CONN, losing 10% on 2P% results in 4.2 less made 2-pt shots (-8.4 pts per game) and losing 5% on 3P% results in 1.4 less made 3-pt shots (-4.2 pts per game) for a whopping total decline of 12.6 pts per game. For a team that was blowing out their opponents on average by 25 pts per game, poorer shooting cut that margin in half.
5. Since opponent shot quantity improved due to fewer turnovers, let's see where those shots went. With an additional 2.2 shots per game, opponents took 1.4 fewer shots from 2-pt land and 3.6 more shots from 3-pt land. This didn't hurt CONN as bad as it could have because opponent 3P% actually fell 0.6%. This is probably the lone bright spot in CONNs last seven games.
6. The improvement in 2P% resulted in an additional 3.5 2-pt shots made per game (+7.0 pts per game) and the increase in shot quantity of 3PA resulted in an additional 0.8 3-pt shots made per game (+2.4 pts per game), which combine for a grand total of +9.4 opponent pts per game. 
7. If NOVA had taken more 2PA (29, least among last 7 opponents) and less 3PA (22, most among last 7 opponents), CONN could be 1-6 in the last 7 games instead of 2-5.
   
8. The -12.6 pts per game for CONN and the +9.4 pts per game for their opponents take CONN's 25.0 margin of victory down to just 3.0. This is why I say throughout my blog that shooting matters. Even taking more 3PA with a marginal decline in 3P% can result in more points per game, which is a big reason among many as for why I hate the 3-pt shot in basketball.

So, how do we explain the rest of the decline in CONN? Rebounds and Free Throws are the only two remaining factors, and (spoiler alert) it is mostly the latter. Let's see what we've got.

Since I've already spoiled the results, I'll briefly talk about rebounds since they correlate more to the aforementioned shooting numbers, and then I'll delve into the free throws.

1. ORBs are up for CONN by 0.8, which is essentially an additional shot per game. It's easy to see why CONN shot quantity didn't waver when possessions per game declined.
2. ORBs are up even more for opponents by 1.6, which is additional two shots per game. When opponent's don't turn the ball over, they get more shots. When they get more shots, they can potentially get more offensive rebounds. It all starts with not turning the ball over.
3. Though its FT% average has remained approximately unchanged, CONN game-by-game FT% has been all over the place, especially when CONN gets less 12 or less FTA. On the contrary, CONN cannot control its opponent's FT%, and it has increased 8.4%.
   
4. As I stated in a previous article, more FTA is better for efficiency than improved FT%. CONN FTA are down 1 per game, resulting in 0.6 less FTM and 0.6 less pts per game. Opponent FTA are up 5.5 per game, resulting in 5.7 more FTM and 5.7 more pts per game. The free throw line accounts for the final 6.3 decline in the margin of victory from 3.0 to -3.3.
5. To see from where all of the free throws came, let's look at fouls. CONN is committing 3.2 more fouls per game whereas their opponents are committing 1.4 less. 
6. I created a simple metric called foul rate (in red), which calculates how many opponent FTAs are generated by each CONN foul. (NOTE: Do not confuse this metric with Dean Oliver's Free Throw Rate or my Free Throw Advantage from the linked article above.) CONN's foul rate has slightly increased from 1.15 to 1.23 (table rounds to one decimal). Even if it stays the same, it is still bad for efficiency if you are committing more fouls per game like CONN is doing.

How do you go from a 25.0 margin of victory to -3.3? The answer to that riddle is 'crashing': Doing all the good things worse and all the bad things consistently. Shooting 10% worse from 3PA and 5% worse from 2PA, defending the 2PA worse by 10%, giving the opponent more shots via more ORBs and less TOs, and giving more FTAs to better FT% teams is a easy way to crash your efficiency numbers. I'm not sure how much of this is CONN and how much of this is CONN quality of opponent, but I would assume CONN should be able to fix a lot of the variables they can control, such as fouling, defensive rebounding, and turnovers forced. They did a lot of this on Sunday (before I wrote this article) against a clearly out-matched BUT team (see the previous match-up against them).

As I mentioned in the "To my readers" section, I can do a lot of these. Honestly, if I had thought about this idea earlier, I could have been doing these weekly following the January QC article. Anyways, thanks for reading my work, and I'll try to figure out something between now and the February QC article.