Bumped from the Fanposts. Should Tony Parker get more mins than he is currently getting under Ben Howland's rotation? Taking a look at the numbers. - BN Eds.
Discussion in a previous thread - specifically on the much-discussed "2-Wear" lineup - led ClassOf66 and I to throw some stats back and forth. In particular, what to make of Kyle Anderson: how to weigh (for example) the fact that he is the best rebounder in our lineup against the fact that he is also our worst shooter.
Many people will have heard of John Hollinger's "Player Efficiency Rating" for the NBA, which provides a composite score for players that attempts to combine value from their scoring, rebounding, assisting, etc, while penalizing them for turnovers, fouls, etc. He provides different weights to different categories in a very complex formula [note: these weights are calibrated for NBA basketball not for NCAA basketball, an issue with what I am about to do...].
The season totals so far for scholarship players are:
|Larry Drew II||445||33||74||7||10||6||21||79||5||34||39||112||22||14||3||21|
From those raw stats, we can get a number of factors that add up to a measure of player efficiency.
|Larry Drew II||1||17||12||1||-5||-6||-0||2||1||3||0||-2||25|
(note: for presentation, I divide all factors by minutes played, and multiply by 100 to get rid of some decimals; this order of operations isn't quite how Hollinger does it, but is mathematically equivalent and makes it easier to see how the factors add up to the whole)
(note2: Tyler Lamb should probably be discounted due to playing only 14 minutes)
Rather than give you the formula for each factor here, I will explain what each tries to capture. (Brave souls can check out the PER formula on Wikipedia. )
A: Value of 3pt shots
B: Value of Assists
C: Value of FGs generally (taking into account how often players on the team score unassisted, ratio of the player's assists to FGs and the ratio of the player's FTs to FGs - essentially, FGs are weighted more highly if the team tends to have a lower assist rate, the player has a high assist rate and tends to draw fouls when he shoots)
D: Value of FTs (weighted slightly less when team assist rate is high)
E,F,G: Penalty for TOs, missed FGs, missed FTs (functions of how likely it is for opponents to get ball, and likely points scored on an average possession)
H, I: Value of Defensive Rebounds, Offensive Rebounds
J, K: Value of Steals, Blocked Shots
L: Penalty for Fouls
For further ease of interpretation, I condense categories further, to show what some common combinations contribute to the overall efficiency rankings:
|Larry Drew II||7||12||1||3||4||-2||25|
What to make of this? Some thoughts in no particular order:
1) Per my initial interest in David Wear and Kyle Anderson as men who play the 4: both have a similar total impact, but by very different routes. David scores a lot more effectively than Kyle (18 vs 7), while Kyle does a little bit of everything else.
2) Of our 7 players who play the bulk of the minutes, Larry Drew II and Norman Powell come out at the bottom; both are significantly worse than Anderson and DWear.
3) Larry Drew's lack of shooting is largely offset by his assists; but he doesn't rebound, get to the line, or win extra possessions with steals or blocks.
4) Norman Powell's shooting is ok, but he rebounds poorly and doesn't get to the line. (it should be noted here that Hollinger's PER cannot capture useful defensive plays that do not result in steals/blocks/defensive rebounds, so it is possible Powell is underrated for that reason)
5) While all three centers are fairly close in PER, Travis Wear is 3rd of the 3 on the initial roster. This would argue for more minutes for Tony Parker, unless you think he is a worse defender than Travis (or can't handle extra minutes). He definitely scores more efficiently, perhaps because he is less enamored of the long, low-percentage 2pt shots...
6) Adams has a bit of a lead on Shabazz at the top of the standings, and both are way above the next regular players. Both win a few extra possessions (Shabazz on the O-boards, Adams with steals), both score in high volume with decent efficiency and make a lot of FTs; but Adams has the edge in protecting the ball and finding open teammates, areas in which Shabazz needs to improve.
In terms of lineups moving forward, I think the PER numbers suggest
a) Tony Parker should receive more minutes (to test whether he can sustain higher performance than Travis in more minutes - and perhaps allow Travis to play more efficiently when better rested)
b) LDII and Powell need to do more and build on what they already contribute, by rebounding better or getting to the line more often - or risk losing minutes to the 3 frosh guards
c) David Wear contributes well enough to justify his minutes (which will result in 2-Wear lineups unless Parker plays 20+ per game) - he is basically as efficient as Anderson but is receiving considerably fewer minutes. According to PER, there seems little evidence of Howland playing favorites with David Wear - at least, in the direction commonly supposed.