The Consistent Stat

With growing interest in removing the soul quantifying basketball, the NBA has done some cool stuff like making plus/minus and "efficiency ratings" available. Pretty cool, but if you've ever had to take an intro stats class, you may have wondered why some basic and potentially insightful statistics are left out of boxscores. Among these, standard deviation sticks out as a good candidate to consider. If you need a refresher, this is a commonly used measure of the dispersion of data around the mean of that data. Or even easier, it tells you how much variation there is in the information you are looking at, a measure of consistency.

Seems useful, right? Would you rather have a player who averages 15 ppg that gives you exactly 15 points every night, or a player who averages 15 ppg but could give you anything between 0 and 30? Averages can miss how consistent a player is, and thats where standard deviation can help. It isn't perfect, but no stat really is.

You could run this on any stat, but I chose scoring since it enjoys such a glamorous life. I went through the game logs of Blazer rotation players and computed the standard deviation ("SD") of their scoring. I've also run a couple non-Blazer superstars for fun. I used the free web based software available here.

For interpretation purposes, I also put the ratio of points per game to SD. This was just to help compare players a little better, because ones with higher ppg are much more likely to have a higher SD due to their greater potential for scoring different point totals. For example, a player that scores 25 a night will have higher SD than a player that scores 6, just from working on a scale with more possibilities. The ratio between ppg and SD can be thought of as the variance relative to ppg.

I also included the median ppg of each player, which I also think is helpful. The median is the middle point if you lined up the player's points total from every game from lowest to highest. This makes it less influenced by outlier scores than averages. For example, a median higher than the average could show a player had a couple really low-scoring games.

Sorry for the long walk, who's the most consistent Blazer when it comes to scoring? Who's the least? I've arranged them in order from highest ppg ("Avg") to SD ratio to lowest:

Brandon Roy:
Avg: 19.23
Med: 21.5
SD: 7.08
Avg/SD: 2.72

LaMarcus Aldridge:
Avg: 17.48
Med: 18
SD: 7.23
Avg/SD: 2.42

Travis Outlaw:
Avg: 11.93
Med: 11.00
SD: 6.15
Avg/SD: 1.94

Martell Webster:
Avg: 11.00
Med: 11.00
SD: 6.23
Avg/SD: 1.77

Channing Frye:
Avg: 7.26
Med: 6
SD: 4.16
Avg/SD: 1.75

Jarrett Jack:
Avg: 9.87
Med: 9.00
SD: 5.95
Avg/SD: 1.66

James Jones:
Avg: 9.78
Med: 11
SD: 5.90
Avg/SD: 1.66

Steve Blake:
Avg: 7.47
Med: 6.50
SD: 4.83
Avg/SD: 1.55

Joel Pryzbilla:
Avg: 4.93
Med: 5.00
SD: 3.52
Avg/SD: 1.40

Sergio Rodriguez:
Avg: 2.86
Med: 2
SD: 2.75
Avg/SD: 1.04

To recap, the higher the Avg/SD ratio, the lower the SD relative to points per game. Higher ratios indicate more consistency.

Any surprises? The biggest might be Blake, but remember he's not really asked to do much scoring beyond what the defense gives him. That could mean a bevvy of open 3's or nothing, depending on the strategy the defense adopts, which in turn leads to lots of variance. Another note is that I rounded to two decimal places, and Jack is indeed a smidge ahead of Jones so gets the higher spot despite the equal rounded scores. These numbers include games completed through 12/31.

And here's some other guys, since I guess its sort of possible you aren't all only interested in only the Trailblazers. You can easily do this for whatever players you want on whatever stat is tracked, just fill out the spreadsheet that pops up on the link to the software used above and have at it. Data entry is where amazing happens.

Kobe Bryant:
Avg: 27.13
Med: 28
SD: 7.08
Avg/SD: 3.83

LeBron James:
Avg: 28.81
Med: 28
SD: 8.76
Avg/SD: 3.29

Yao Ming:
Avg: 21.5
Med: 22
SD: 6.63
Avg/SD: 3.24

Dwyane Wade:
Avg: 24.79
Med: 23.50
SD: 9.80
Avg/SD: 2.53