Monday, April 12, 2010

Size really does matter

I've always been a fan of size in basketball. It's hard not to be. Taller teams - as long as they don't sacrifice too much skill or quickness - tend to be better defending and rebounding teams. Not to mention that effective low post play is one of the best ways, if not the best way, to score points efficiently.

Of course, the Hoyas' tendency has been to play small.

That's not to say that the Hoyas are small. They aren't. Traditionally, the personnel has always been tall. Even under JTIII, the Hoyas have been tall.

But they've still played small.

For example, the Hoyas were 12th in effective height this year. But they played small far too often. What do I mean by that?

They were "tall" because Monroe was 6'11" and played a lot of minutes, and didn't have a sub-6' guard. But of the roughly 1,300 possessions the Hoyas played that mattered from the Washington game onward, 88% of them were with a lineup "smaller" than the traditional two guards, wing and two bigs.

I took Brian's excellent lineup data for the regular season and designated each lineup by type. I classified Freeman, Wright, Clark and Sanford as guards, Thompson and Benimon as wings, and Vaughn, Sims and Monroe as bigs.

Here's the results in terms of efficiency, by type of lineup:
Lineup Type      WAVG OE       WAVG DE      DIFF

2G 1W 2B           117            93         +23
2G 2W 1B            96           105          -9
3G 0W 2B           113           106          +8
3G 1W 1B           108           100          +8
Lineup Type is designated with the number of positions then the position; in other words, the first line is "2G 1W 2B", which is read as "2 guards, 1 wing, and 2 bigs."  
WAVG OE is the Weight Average Offensive Efficiency;  
WAVG DE is the same for Defensive Efficiency. "Diff" is the difference between the two.

Some thoughts:
  • I used weighted averages to avoid having lineups with small sample sizes affecting the numbers. It's necessary given how few lineups the Hoyas run out there for substantial amounts of possessions and also helps negate some of the competitive issues. That said, keep in mind that a lot of Hoyas played a lot of minutes -- that 1B that is so prevalent is predominantly Monroe, for example.
  • The biggest caveat here is that this does not adjust for quality of player. That's okay, because I'm not trying to prove larger lineups are better in general; I'm just trying to examine if a larger lineup will work for the Hoyas.
  • That last point is apparent by seeing the results for the "2G2W1B" lineup and how poorly it performed. I don't think that the awful offense and poor defense were driven by the size of the lineup. This is just a situation where both Jerelle Benimon and Hollis Thompson were out on the floor together. Two freshmen, one a complete offensive liability, is not a great combination for any team. What is bad is that that particular duo played together for 186 possessions, more than the traditional 2G1W2B lineup.
  • The three guard lineups played 74% of all possessions.
  • The best lineup on both defense and offense was the traditional 2 guard, 1 wing and two bigs lineup. This was the least used lineup type.
  • There were only four versions of the traditional and most effective lineup type. Hollis Thompson was the small forward in all of them. The bigs rotated between Vaughn and Monroe and Sims and Monroe. Freeman, Clark and Wright are in three of the four each, but the most common lineups were Monroe-Vaughn-Thompson-Freeman with Wright or Clark at the point.
  • All four of those lineups held opponents under 100 efficiency. Three of the four (and both of the most common) were at 110 or above offensive efficiency.
  • The three guard lineups were similarly performing, though they went at it different ways. Oddly, the two bigs lineup was better offensively, but weaker defensively. This likely speaks to matchups.

Analysis like this is far from definitive. There are reasons for playing small that would get lost at this level of detail. Certain lineups were less effective overall, but that doesn't necessarily mean the more effective lineups would have been more effective in all situations. Sometimes the most effective lineups or players only are so because they are limited to situations where they will be so effective.

For example, many of the small lineups could have been played against opponents with power forwards that were perhaps too quick for Julian Vaughn. Perhaps if Thompson had gone bigger, the result would not have been better.

That said, the results certainly raise the question, don't they? I think all three guards were better players than Hollis Thompson or Julian Vaughn this year, overall. I'd be willing to bet Thompson agrees. But that doesn't mean that they are necessarily in the best five. Sometimes you just need size. The numbers seem to support that.

Thompson seemingly loves to play a little down, perhaps because he puts such a premium on skill. I don't use the word "small" here because it isn't so much actual size as type of play. He started Brandon Bowman at PF for a year, DaJuan Summers for two, and both of those guys were small forwards at heart (though Brandon took to PF more than Summers did), despite being 6'8". Last year, a PF started, but Vaughn only played about half the minutes, so for the rest of the time, the Hoyas had a SF at PF.

Aside from 05-06 and 06-07 (our most successful tournament runs, by the way), JTIII's Hoyas never really had two "bigs" on the floor at all times.

How much of that is personnel and depth? How much is preference? Next year the Hoyas will likely be again looking at a situation where the guards are further along than the bigs. Will we see a repeat of small lineups? There's an argument to be made for playing real small forwards and real bigs despite the fact that they may not be as good as the guards from whom they'd be taking minutes.

And they would be taking minutes. The four Hoyas I established as guards (Freeman, Wright, Clark and Sanford) took up 2.67 positions worth of time last year. They'll be adding Markel Starks to the mix as well.

Playing a normal lineup all the time means cutting more than most people would want to. Assuming a coach wants Vee and Markel to play some, that means cutting close to 30 minutes out of Freeman, Wright and Clark's PT to get the team to a traditional lineup for all forty minutes.

Anyone want to do that?

Didn't think so. Even the most ardent "Big Lineup" supporter would be looking to play a 3 guard lineup at least 10-15 minutes a game, leaving the team to cut about five minutes from Clark, Freeman and Wright apiece.

Does the added benefit of height and size of outweigh sitting Georgetown's better players? It seems that this year it might have. The traditional lineups simply performed better than the three guard lineups. If everyone returns, why wouldn't that be true next year?

Hollis will likely improve more than the guards will, given that he's in that freshman to sophomore improvement sweet spot. And one of the team's biggest issues with playing two bigs at once this year was the team's lack of effective bigs.

But Sims showed some strides at the end of the year, and both he and Vaughn will have a summer to shore up weaknesses. Throw in Lubick and Abraham and the team seems to be better equipped to play two bigs at once next year than they were in 09-10.

That said, everything changes if Monroe leaves. While the team size advantage could offset the individual effectiveness difference between say, Clark and Hollis or Clark and Vaughn, I don't think that equation works with Clark and a freshman or Sims.

In other words, get used to a smaller lineup. There's going to be substantial three guard lineups next year, no matter what, simply because it seems unlikely any of our big three guards are going to see ten minute drops in PT.

But if Greg stays, there's a strong argument that Coach Thompson should play two guards, a wing, and two bigs as much as possible, even if those aren't necessarily the five best players.