Monday, January 28, 2013

How to Beat Louisville

On January 18th, the Louisville Cardinals sat at 16-1 (4-0 in the Big East) and number one in the country. Louisville featured the number one ranked defense (efficiency-wise) in the country and a greatly improved offense from 2011-12. Now, 10 days later, the Cardinals are 4-3 in the Big East with losses to Syracuse, Villanova, and Georgetown.

In the final minute of Louisville's first loss (in the streak) to Syracuse, it was a Louisville turnover leading to a Michael Carter-Williams dunk that propelled Syracuse to a victory. I immediately thought that had to be the most costly turnover in a single game to date. Not only did Louisville lose the ability to score any points of their own on that possession, but they gave Carter-Williams a free fastbreak to take the lead. All turnovers are not created equally. Live ball turnovers lead to fastbreaks (and thus a higher expected point value for the team forcing the turnover), while dead ball turnovers simply lead to an offensive possession similar to when a shot is made. Going even further, Louisville turnovers are even more valuable for their opponents, because you get to skip the daunting task a beating Pitino's fullcourt press.

Using the numbers at, we can see how preventing Louisville from scoring impacts BOTH defensive and offensive efficiency:

After Louisville makes, the press has forced teams into an eFG% of just 40% (0.8 points per shot). When Louisville does not make a shot on offense, their next defensive possession has been significantly affected. Opponents are scoring over 1 point per shot in those scenarios. Furthermore, this does not take into account turnovers, which are presumably higher against the press than otherwise.

I broke down Syracuse's offensive possessions in a similar manner. The sample size here is obviously smaller, but the results are in more detail:

*I did not look at anything past the initial shot, offensive rebounds (and the 2nd shot attempts that comes with them) were excluded from the data

Key Takeaways

Louisville's press was lethal. Of the 27 times Louisville was able to press, Syracuse scored just .63 points per initial play (which excludes second chance points). Without the press, Louisville allowed a very human 1.11 points per initial play (which excludes second chance points).

Syracuse's guards played a bigger role in the offense when Louisville pressed. Pretty obvious point here, but the data backs it up very nicely (see the orange chart).

Syracuse was unable to get to the rim frequently against the press. This one surprised me, but Syracuse took a lot of three against the press and not a lot of the threes against the halfcourt match-up zone (counterintuitive).

Friday, January 4, 2013

Defending the Free Throw

Recently, there has been a lot of talk about foul shooting and luck. The discussion has been about the luck the shooter has during a foul shot. This is very controversial, because we as humans like to believe we are in full control  over everything. However, I don't think anyone would dispute the luck involved in foul shooting for the "defense". Using the word defense for a FREE throw by definition doesn't make sense. The defense simply watches while the opposition shoots.

St. Mary's is 11-3 this year, most recently with a comeback win over Harvard at home. They have shot the ball extremely well this season (4th in eFG%), but have also let their opponent's shoot well. The Gaels are 326th in defensive 3P% and 320th in defensive FT%. There has been much talk about opponent's 3P% being misleading and a lottery, but the same obviously goes for FT%. 

The following are graph of St. Mary's 13 D1 games this season. The graphs plots their opponent's 3P% and FT% in the game against St. Mary's relative to their season averages (average % - St. Mary's game %). Basically, if a data point is above the x-axis that means the Gaels were lucky in their game and if a data point is below the x-axis that means the Gaels were unlucky.

St. Mary's luck (or lack thereof) has very likely just been due to random variation. I would bet that their defense will improve by the end of the season simply from regression to the mean. Still, I came up with three ideas (I'm sure there are more) that would possibly affect opponent's FT%:

1) Tempo/Fatigue - It's possible a team could play at a level of speed or physicality high enough to influence their opponent's ability to shoot. 

2) Location/Game Environment - Crowd, travel, familiarity, etc. could all lead to bad shooting.

3) Player Being Fouled - Guards generally shoot better than big guys. If a defense had guards who were foul-prone, it's safe to say teams would shoot better from the line.

Next Up: Teams expected to gain from unlucky shooting