Tuesday, March 27, 2012

2012 Final Four Preview: Is the "Trifecta" a bad bet for Kentucky and Louisville?

I get distracted a lot, mostly by sports-related topics that non-sports people would find boring (the one-and-done rule, the NFL's tax exemption as a non-profit, the various ways to throw a curveball, etc.). On some occasions, I find myself to have a sudden fascination with numbers, though I am wretched when it comes to keeping track of figures.  However, my lack of arithmetic skills aside, I have a certain curiosity about efficiency and the ways in which things are done. I attended a lecture by Moneyball guru Paul DePodesta a few weeks ago, and he emphasized the importance of asking what he termed the naive question (which he credited to a man named Peter Drucker):

"If we weren't already doing it this way, is this the way we would start?"

Hitting "trifectas" is usually a jackpot, but shooting
too many could spell disaster for Kentucky or Louisville.
So I thought about the three-point shot, specifically about how it figures into an offensive gameplan. The three-pointer has become the cliched "double-edged" sword of college basketball, particularly in March when teams are trying to rally from big late-game deficits in the NCAA tournament. If you went back and looked at every preview box for the 68 teams on the ESPN and Yahoo! Sports bracket challenge games, I swear you'd find at least 41.2657% of the analysis done on the team's chances included a mention of their prowess or deficiency in shooting from beyond the arc.  On its face, an offense built on knocking down 3s makes some sense. True, it's a lower percentage shot than your run-of-the-mill two pointer, but it's great for teams in need of a comeback, assuming they shoot it well. Example:

  • Situation: Comeback Kids are down by 12, four minutes to play, possessing the ball
    • Two-point solution: The Kids, who favor two-pointers, have to make six FGs (with almost no fouls committed by the Leaders) as quickly as possible AND play stellar defense a minimum of five times just to tie the game. At best, if they score in the first ten seconds of each possession and can't manage a steal but keep the Leaders from scoring, the Kids will get the ball with 15 seconds to play needing to score to tie the game. If they don't, they lose.
    • Three-point solution: The Kids, now favoring three-pointers, only have to make four consecutive shots (remember, the Leaders will mostly avoid fouling) and get three defensive stops in order to tie the game. Not only do the Kids have more time to mount a rally - more importantly, they have to do less in that time than if they were all shooting jumpers. At worst, the Kids can use their entire shot clock, defend without trying to steal the ball and still get possession with 30 seconds to play needing a three to tie the game. The best case scenario? If they don't steal but are still able to score in the first ten seconds of each possession, they'll tie the game with 1:45 to play and are certain to get the ball back, with a chance to win, at least two more times if they defend and rebound properly. 

I'm sure someone has come up with a stat to demonstrate the usefulness of a trey-centric offense (please point it out if you know it), but I wanted to run my own numbers for the sake of procrastination and satisfying my need to feel like some strange intrepid fan-sleuth.

Using TeamRankings.com, I looked up a bunch of the team shooting and scoring stats from the past 4-5 years and tried to determine a few things:
  • In a 40 minute game with a 35 second shot clock, there are about 68 possessions if every shot clock is used fully, giving each team about 34 shots in a given game. HOWEVER, teams often take very early shots, especially with the prevalence of fast-break offenses and man-to-man defenses now-a-days. Consider:
  • The average NCAA Division 1 men's basketball team shoots about 55 times per game from the floor.
  • The average team FG percentage is roughly 42.5%.
  • That said, keep in mind that the average team also shoots about 20 3-point shots per game. (Sounds high, right? It is, a little bit - the real figure is about 18 per game. But these calculations are pretty rough, gimme some leeway for a moment.)
  • On those three-point shots, the average team hits about 35%.

20*.35 = 7 three-pointers made per game = 21 points from 3-land
55-20 = 35 two-point attempts per game

In order for the overall team FG% to be around 42.5%, the % from two-land has to be slightly higher to balance out the 35%, right? Sure. After a few trials, I found 46.5% to be a valid input here. (Not surprisingly, the actual 2pt percentage for average teams is about 47%).

Therefore, 35*.465 = 16 two-pointers made per game = 32 points the old-fashioned way

If at this point you think that 3-pointers are not as valuable as 2-pointers, you'd be right where I was about twenty minutes ago. Then I thought about the discrepancy between the number of attempts, so I looked at my numbers in terms of points per shot:
That's the spirit, Thad! But be
careful not to shoot too many.
21 points scored on 20 three-point shots
32 points scored on 35 two-point shots

Aha! Now we have a mismatch: it seems that long-range bombs are about 15% more efficient in the long run than the good ol' jumper. I think  the next logical thing to examine would be whether or not a law of diminishing returns can be applied, i.e. if shooting more than a certain number of three-point shots makes the efficacy of the shot diminish, as on a bell-curve.

Food for thought (scoring composition in brief):
  • Kentucky: 77.9 ppg, 17.1 ppg off 3s, 43.6 ppg off 2-pointers      
  • Louisville: 68.4 ppg,  17.6 PO3/g, 36.4 PO2/g
  • Kansas: 74.2 ppg, 17.4 PO3/g, 41.2 PO2/g
  • Ohio State: 75 ppg, 15.1 PO3/g, 39.1 PO2/g
Now we get to investigate the law of diminishing returns as it applies to the three-point shot. We already guessed that Average U. takes about 18-20 three-pointers per game. Virginia Military Institute (VMI) has been known in recent years for launching an absurd amount of triples: the Keydets have averaged more than 26 attempts per game in each of the last six years, registering as high as 40 in 2006-2007. However, they haven't made the NCAA tournament even once during that time, another proof that extremism in any form is generally not a good thing.

Of the teams in the last four Final Fours, here's where they ranked in 3pt attempts:

2012                    3PA/G         Rank
Kentucky  ?             15.1          288
Louisville  ?           18.5          143
Kansas     ?          16.8          217
Ohio State ?          15.1          286

2011                    3PA         Rank
UConn*               17.5          195
Butler                 21.2          52
Kentucky            18.6          145
VCU                    22.9          20

2010                    3PA         Rank
Duke*                    19.6          105
Butler                    20.0          81
WVU                    20.1          80
Michigan State     14.7          306

2009                    3PA         Rank
UNC*                    17.8          175
Nova                    18.4          149
UConn                  13.4        325
Michigan State      14.9          296    

2008                    3PA          Rank
Kansas*               16.9          250
Memphis               21.6          68
UNC                    14.9          313
UCLA                    15.6          301


As the stats show, the teams from the non-major conferences - VCU, Memphis and Butler (twice) - relied on the long ball to reach the Final Four, but all of the winning teams shot fewer than 20 three-pointers on average.  The only team since 1998 to win it all using the three-pointer as a dedicated weapon was the 2001 Duke Blue Devils, who shot 26 threes per game and were officially an offensive anomaly with four starters - Jay Williams (42.7%), Shane Battier (41.9%),  Mike Dunleavy (37.3%), and Chris Duhon (36.1) - shooting lasers from three-range and averaging more than 27 points per game combined on deep shots alone. [With the support of Battier and Williams sinking 256 combined treys, the team set records for most three-pointers made (407) and attempted (1,057) in a single season.] The only other recent year any team has even had two staters in that neighborhood was the following season, when Mizzou's Clarence Gilbert (38.4%) and Kareem Rush (41.6%) combined to average over 18 ppg on three-balls.

And from all this madness, I've learned...what? Well, it seems like putting up more than an average number of 3s per game is a bad idea unless you have the most talented shooters in the world, though I suppose that's somewhat self-evident. Anybody wanna guess how many 3-pointers Kentucky put up in their last loss? That'd be 28. And Louisville? 23 attempts in their most recent loss to Syracuse.  Interestingly enough, Ohio State and Kansas have not surpassed 20 three-point attempts in any of their last three respective losses. Might a deep-ball drought spell doom for the Bluegrass teams, given Louisville's comparative lack of offensive firepower & Kentucky's method of being judicious from distance? You could draw that conclusion & rationally decide that picking the winner of the Kansas-OSU game might be a safer bet**. Or you might make that pick based just on what color you flipped a coin for, red or blue. Me? I'm just nuts enough take the Buckeyes.

**(Wait: does this mean extremism as it applies to efficiency - either really inefficient or totally effective - is a bad thing? Well When there's a bias in favor of balance, as I have consciously or subconsciously identified here, even good efficiency in the long-term is dangerous, because nobody is safe from "one bad night.")