LeBron James is the most consistent of the NBA's top scorers. What's that worth? Who are the most volatile scorers in the league?
Feb 2, 2012 - As we know based on the wide range of shooting percentages we see among the list of the NBA's top point producers, we know that not all scorers are created equal. Kobe Bryant is scoring a bit more than LeBron James per game this season, but Kobe needs four extra shooting possessions to reach that level. Carmelo Anthony is the league's No. 5 scorer, but takes more shots per game than two of the guys (Kevin Durant, Kevin Love) ahead of him on the list.
There's another way in which scorers differ: volatility. Some scorers are more reliable than others; there are the proverbial "you know you're going to get 20 points" guys and the "explosive scorers." We sometimes give credit to explosive scorers, even those in prominent offensive roles, because they could go off at any time and get their team back in the game.
But there's a flip side to that: they aren't necessarily dependable. In the NBA as in any competitive atmosphere, consistency has some importance. If a team's management doesn't know what they'll get out of a star any given night, it's hard to build a proper cast around him. Just how much of a drain excessive volatility is a question mark, at least for this column. (I have no doubt a team or three have conducted studies on the value of consistency, but there's nothing public or publicized in the NBA realm on that issue that I know of.) This exercise is meant to look at which players are volatile compared to their peers.
There are currently 15 players who have enough minutes to qualify for the leaderboards who are averaging 20 points per game this season, ranging from Dwight Howard's 20.1 average to Kobe's 30.0. To measure their scoring volatility, I borrowed a trick from finance: I took the standard deviations of the players' game-by-game scoring totals and divided by their scoring average. This gives us a volatility rating that allows us to compare 30-point scorers with 20-point scorers without bias.
| Player | Volatility |
|---|---|
| Carmelo Anthony | 44.0% |
| Dwight Howard | 41.5% |
| Monta Ellis | 40.9% |
| Brandon Jennings | 40.6% |
| Kevin Martin | 38.6% |
| Chris Bosh | 38.4% |
| Deron Williams | 37.2% |
| Russell Westbrook | 35.7% |
| Derrick Rose | 32.7% |
| Kobe Bryant | 29.6% |
| Kevin Love | 27.2% |
| Blake Griffin | 25.7% |
| LaMarcus Aldridge | 23.7% |
| Kevin Durant | 23.3% |
| LeBron James | 20.8% |
What we found is not wholly surprising: 'Melo, who this week actually had a 1-point game, is the most volatile of the 20-point scorers this season, and not by a small margin. Anthony's game-by-game standard deviation is nearly half the size of his scoring average; he is almost as likely to have an extreme scoring performance (more than 34 percent above or below his average) as he is to have a normal one.
Howard, surprisingly, is right behind 'Melo. Seeing a big man so high on the list fights conventional wisdom, which is that since big men tend to have higher shooting percentages, their production is more reliable. While Kevin Love, Blake Griffin and LaMarcus Aldridge all seem to bolster that hypothesis with relatively low volatility marks, Howard -- a regular near the top of field goal percentage rankings -- is a blinking signal of doubt.
One possible explanation: Dwight earns so many free throw attempts and shoots them so poorly that if he has a bad day at the stripe, he's likely to have a bad day in the scoring column. (That the season has been short and that he had a monster scoring game against Golden State could affect things too, though the game didn't throw his average too far away from expectations.)
The other numbers aren't terribly surprising. LeBron is a clock; he's had just two games under 20 points in 21 outings and none under 16. Kevin Durant is similarly stable, with the second lowest volatility rating among the 20-point scorers. Chris Bosh is, other than Howard, the most volatile of the high-scoring big man, likely owing to his rare role as a third option on offense. (Not many third options average 20 a game.) Monta Ellis and Brandon Jennings live up to their reputations as "explosive," and Russell Westbrook looks more like Derrick Rose every day.
Again, we don't know how important volatility in scoring really is, so it's hard to place value on the predictability LeBron, Durant and most of the big men provide. But it's worth keeping an eye on as we suss out our expectations for what players can provide and how consistently they should do so.
The Hook runs Monday through Friday. See the archives.
10 comments
Comments
I’d argue we don’t the importance of reiability to good teams but we have a pretty good idea on bad teams. If Ellis doesn’t explode, the Warriors can’t compete against good teams and perhaps lose to an even-leveled team. Same with Martin and even Jennings, to some extent.
Rafael Uehara
by Rafael.Uehara on Feb 2, 2012 11:00 AM EST reply actions
Point evolution since 2006
Consistency is by far the best quality of an athlete regardless of the sport. The fans & coaching staff gets what they pay for. Having a consistent superstar is a priceless asset for a team. Here’s a graph representing the point scored in the last 6 years by LeBron James, Kobe Bryant, Chris Bosh & Kevin Durant, all of them seems consistent (maybe not Bosh though).
http://sportsnetics.com/s?id=79
by DougD1 on Feb 2, 2012 11:49 AM EST reply actions
Sure, if they are consistently great. What if they are consistently terrible?
The question is if you have two guys, who have the same mean stats, which one is more valuable, the volatile one or the consistent one? It may actually depend on the team. A volatile scorer on a bad team may get you some wins because he has a few outstanding games, whereas the consistent player never has a performance that lifts his bad team up enough. On a good team, though, a consistent player might be more valuable. I suppose one could simulate such scenarios, but I don’t think it’s trivial.
Follow @thecity2
Read my Advanced Stats Primer
J-RIDAH: Its not 1 player in this draft better than Monta or Lee. Anthony Davis is no different than Al Farouq Aminu. Andre Drummond could be good but he is not impressive at this point at all besides his size. This draft is hella overated.
(JaVale) Mcgee is better than anybody in this draft.
by Evanz on Feb 2, 2012 12:39 PM EST up reply actions
Is this technically called “volatility”? The other day I used exactly this calculation, but I wasn’t sure what to call it. It’s like a normalized standard dev?
If you standardized each player’s totals and then took the standard dev of that, would that be equivalent?
Follow @thecity2
Read my Advanced Stats Primer
J-RIDAH: Its not 1 player in this draft better than Monta or Lee. Anthony Davis is no different than Al Farouq Aminu. Andre Drummond could be good but he is not impressive at this point at all besides his size. This draft is hella overated.
(JaVale) Mcgee is better than anybody in this draft.
by Evanz on Feb 2, 2012 12:14 PM EST reply actions
I also wonder if this falls under DeanO’s “risky strategy” discussion in BoP. Do teams with less talent need to be more volatile to win?
Follow @thecity2
Read my Advanced Stats Primer
J-RIDAH: Its not 1 player in this draft better than Monta or Lee. Anthony Davis is no different than Al Farouq Aminu. Andre Drummond could be good but he is not impressive at this point at all besides his size. This draft is hella overated.
(JaVale) Mcgee is better than anybody in this draft.
by Evanz on Feb 2, 2012 12:16 PM EST reply actions
They certainly are encouraged to take more risks.
It’s a reason you see bad college football teams attempt more (unexpected) onside kicks and go for it more often on fourth downs, like Duke’s David Cutcliffe did to pretty good effect this year.
I gotta stop takin' my baths durin' Peter's shenanigans.
by MichaelProcton on Feb 2, 2012 10:35 PM EST up reply actions
I'm not a big stats guy - but I like this one a lot
I’m uncertain whether how it’s ultimately being expressed here though translates to the common fan.
I think we need a mean and an average to use as a baseline(?)
But I think this is one you should not only push but find a way to translate into other starters who average say 15 or more or add minutes as in a ‘per’ formula.
"Victory goes to the player who makes the next-to-last mistake."
- Chessmaster Savielly Grigorievitch Tartakower
by lietothegirls on Feb 2, 2012 4:21 PM EST reply actions
Perhaps 'reliability' could be a better expression?
"Victory goes to the player who makes the next-to-last mistake."
- Chessmaster Savielly Grigorievitch Tartakower
by lietothegirls on Feb 2, 2012 4:22 PM EST up reply actions
This analysis is mathematically flawed
The simplest way to model a players scoring is a Poisson Distribution, or maybe a weighted Poisson Distribution. However, the variance, not the standard deviation will be proportional to the mean. The variance is the standard deviation squared, so taking standard deviation/mean will give something proportional to 1/sqrt(mean). Thus, even for players whose scoring is a perfect Poisson Distribution, the higher scoring players will appear more consistent.
by jnewhouse on Feb 2, 2012 10:14 PM EST reply actions
Any chance you can give some more examples?
I’d be curious how the previous, say, 5 NBA finals looked in terms of the two teams "offensive volatility, preferably based off the playoffs if it’s not too small a sample size.
Our Fair City...a campy post-apocalyptic science fiction radio epic!
by The BBQ Chicken Madness on Feb 2, 2012 10:37 PM EST reply actions
Comments For This Post Are Closed