Sunday, April 19, 2009

Season Post-Mortem IV: More "Luck"

A couple of weeks ago, Tom took a look to see if part of the Hoyas' troubles were caused by "bad luck" due to abnormally high opponents' free throw shooting. He found that most of the Hoyas' bad luck came from playing good shooting teams.

Still, it was a good idea to look into "luck." Even at the end of the season, the Hoyas were still considered a good team by many of the statistical systems out there. That's because those systems look at full year points scored and points allowed. In most cases, those numbers predict actual winning percentage very well. When they don't, the system Pomeroy uses, for example, kicks the difference out to a factors he's unfortunately labelled "luck."

The prime driver of this statistic is performance in close games. (Blowouts can also play a role -- teams that blowout awful teams by more than other teams can be a bit overrated.) There's not doubt that actual random chance can play a role in close games. Think of how many times borderline calls have been made in a one point game, for example.

But there's also a whole lot of execution. Using "luck" to describe the fact that someone doesn't step on the end line in a blowout but does in a close game isn't quite right. Actual performance isn't so consistent as to attribute variances to chance; players play better sometimes and not as well other times. You can call that luck, and you may be right, but you certainly have to also leave open the possibility that college players can be affected by the situation.

In other words, did the Hoyas crumble down the stretch in tight games?

Subjectively, it certainly looked like it all year. How do the numbers back it up?

Year         Pomeroy Luck       Games Lucky/(Unlucky)
2005-06 -.035 -1.2
2006-07 .001 0.0
2007-08 .021 0.7
2008-09 -.113 -3.5


The Hoyas lost three to four more games than they should have based on their points scored and points allowed last year. That's an additional three to four losses on top of most of what plagued the team all year -- it's not how much the Hoyas scored allowed, it's when.

So what causes this?

One possibility is this is simply a consequence of the fact that Pomeroy calculates his ratings on a full year basis. If the team really did, in fact, collapse halfway through the season and in effect become a different team, then the early season Hoyas might be holding up the rating and the late season Hoyas might have just stunk instead of playing poorly in close games. We'll check that in a later post.

Another possible cause is youth. I've looked into that here and found that while you can't rule it out, there's not a lot of evidence saying that youth is a common driver of "bad luck," which is somewhat surprising.

Another possibility, though, is that it really just was bad luck. Some shots rimmed out, some fouls weren't called, some opponents' shots went in. If that's true, then that is good news for Hoya fans -- this team may just get a bit better by regressing to the mean in "luck" next year.

Tuesday, April 7, 2009

Free Throw Defense

In looking at the Hoyas' profile on KenPom for the 2008-2009 season, and particularly at the shooting defense, the Hoyas weren't a respectable defensive team in terms of FG% and 3FG%-not up to the standards of the defensively quite good 2007 and 2008 teams, but better than the 2006 squad that reached the Sweet 16. But, Hoya opponents shot 70.8% from the line last year, a well above-average mark and much better than the 68.0% they shot in 2008. The question I wanted to know the answer to is why.

One potential answer is bad luck. The Hoyas had some clear examples of this, particularly losing to Cincinnati in overtime when the Bearcats shot 6-for-6 as opposed to the Hoyas' 4-for-6 in the extra session and Jonny Flynn's 15-16 in the OT loss in the Carrier Dome.

Another potential answer is fouling good free throw shooters. Against Syracuse, fouling Flynn as opposed to Onuaku gives Syracuse an extra almost half an expected point from the line.

To answer these questions, I examined Hoya opponents' free throw shooting beginning with the Big East conference opener, both in terms of how well the team shot compared to how well the team shot over the course of the entire season and how well the individual players shot that game compared to their season statistics. The answer is that the Hoyas were indeed screwed by the luck fairy in terms of opponent free throw shooting, which cost them several comes. Unusually, the luck fairy had an actual name: the St. John's Red Storm. Over the course of those 21 games, Hoya opponents made 10.6 more free throws than you would expect from their season stats, and 10.0 more free throws than you would expect from their players' season stats. The Red Storm were responsible for 9.2 and 9.3 of those, respectively. That means, in those other 19 games, Hoya opponents made 1.4 and 0.7 more free throws than you would have expected based on the team and player stats, respectively.

Here are the numbers for each game, with + numbers meaning the team made more free throws than you would have expected and - numbers meaning fewer made free throws than expected:

Opponent Team Players UConn +1.1 +1.1 Pitt -1.4 -1.0 Notre Dame +3.6 +2.8 Providence -1.2 0.0 Syracuse -4.9 -4.1 Duke +2.5 +2.1 West Va. 0.0 +0.1 Seton Hall -0.3 -0.5 Cincinnati +1.2 +0.2 Marquette +2.3 +2.6 Rutgers +1.3 +1.1 Cincinnati +0.5 +0.1 Syracuse +4.0 +3.8 South Fl. -4.8 -5.3 Marquette -1.1 -0.3 Louisville -0.5 +0.6 Villanova +0.7 +0.2 St. John's +4.0 +4.6 DePaul -1.7 -2.0 SJU (BET) +5.2 +4.7 Baylor +0.1 -0.4 TOTALS +10.6 +10.0
Aside from the aforementioned St. John's games, a couple other games worth noting.
  • Syracuse did indeed shoot better than expected in their overtime win at the Carrier Dome. They were, in fact, about as hot from the charity stripe that night as they were cold in the Hoyas' 88-74 win at home.
  • Cincinnati's perfect-6 in the extra session was not particularly lucky. The Bearcats were only 7 of 13 in regulation- if they shot at their season average, that game doesn't make it to overtime.
  • The only other game where the FT luck advantage is close to the margin of victory is the Notre Dame game - 3 "luck" points in a game that ended with a 6 point margin.
  • The difference over these 21 games between the overall team stats and the per-player stats was 0.6 expected made free throws. I think we can lay to rest the idea that the Hoyas had a problem this past season with fouling their opponents' best free throw shooters.

Bottom line: On the whole, Hoya opponents made a lot of their free throws this season because the teams the Hoyas played were good at making free throws.

Areas for expansion:
1. Figuring out the numbers for the pre-BE season games. I'll add these to this post in a day or three.
2. Figuring out the same numbers for the Hoyas. This will be a separate post, and will include what the result would have been if both teams had shot free throws at their season averages.
3. Figuring out why Hoya opponents made only 68.0% of the free throws last year, when they made 70.8% this year and 71.0% in 2007.

Saturday, April 4, 2009

Season Post-Mortem III: Shot Selection

In Part II of my masochistic rehashing of the season, we looked at the Hoyas' three point shooting and how it all but completely collapsed this season. But ability to make shots is not only affected by shooting ability, it is also affected by what type of shots you take.

How has the Hoyas' shot selection changed over the last three years and what effect has it had?

The Hoyas' offense has declined over the last two years, and its True Shooting % has declined in course. True Shooting % incorporates the added point for three pointers as well as FT shooting.

In 2006-07, the team had a 60.7% TS%. The next year, with Jeff Green gone, it slid to 58.9%. Last year, it slid again, though not nearly as much, to 58.0%.

Thanks to CO's tireless work, we can see the distribution of the Hoyas' shots between dunks, layups, two-point jumpers, threes and free throws. So was this drop because the Hoyas were taking different kinds of shots, or were they taking the same kinds of shots and just not making them?

The reality is, the Hoyas' shooting choices really haven't changed that much, at least by the details that a box scores afford us.

The 2009 shot selection went as follows: 3% on dunks, 32% on layups, 19% on two-point jumpers, 31% on threes and 15% on free throws.

In neither of the previous two years did any of these percentages differ by more than 4% (The Hoyas shot 35% of our shots from three last year). In the Final Four year, the Hoyas had more dunks and less layups, and a few more two point jumpers, but none of these percentages differed by more than 3%.

What's left is simply the ability to make these shots. Compared to last year, the Hoyas were much less effective on dunks and layups, and also significantly less effective on threes.

Compare to 2006-2007, the Hoyas were less effective in both those areas and two point jumpers.

Of course, it's important to note that that one percent drop from last year is equivalent to about twenty points over the course of the season; the three point drop is just sixty points.

Some thoughts:
  • One to two points per game is more significant than you think, but this also speaks to the importance of getting more shots (offensive rebounds, turnovers) and defense.

  • The reduced shooting percentage doesn't mean shot selection isn't an issue. The team still may be taking worse threes and worse layups. But the team isn't settling for too many jumpers, at least compared to the past.

  • In all three years, two point jumpers scored at least at a 20% less points per shot rate than the next worst choice. People trumpet the mid-range game, and there's certainly a place for it. That said, there's a big gap to make up in taking better two point shots before it is really an effective weapon rather than something you settle for.

  • The offense probably looks better than it was because it was a good offense for half the season. When I get a chance, I'll look at Big East numbers.