clock menu more-arrow no yes

Filed under:

Andrew Wiggins and Asking Questions Through Statistics

New, comments

The following was written by Derek Bodner as a comment in this Andrew Wiggins post by Rich Hofmann. It's great, and deserving of its own post, so here that is.

Scott Rovak-USA TODAY Sports

I think traditional scouting is prone to biases. Physique bias, athleticism bias, overvaluing of certain skill sets, underestimating how much certain skills can improve. No matter how much basketball you've watched, we're all human.

I say this, yet I still spend a thousand hours per year watching college basketball. There's a reason for that.

Baseball is the pinnacle of statistical analysis because of its unique ability to limit noise and increase sample size. Baseball, more than any other team sport, is a game of 1-1 matchups. Sure, there's some noise that gets in the way: from opponents to parks and whatnot, but they're fairly well limited because of the sheer number of games they play. Not perfect, but the combination of 1-1 matchups and the sample size is as good as we get in sports.

Andrew Wiggins Takes

College basketball has significantly more noise than the NBA. Discrepancy in age, role, opportunity, coaching style, quality of teammates, quality of opponents, travel, and each individuals basketball age, or how far along they are in their progress. For some of these guys, they're a skill or an understanding away from drastically changing their impact. That's not even talking about how insanely small the sample size is for some of these guys. 35 games? When adjusting to new teammates, new coach, and new rules? We can't even pretend that this is a statistically significant sample to make long term declarations.

While some of these models are better at limiting the noise than others, all have noise, and a high degree of noise creates a high degree of outliers, if you can even call it that. What you get, instead of truths, are guides.

When looking at any statistic, advanced, traditional, or what, the first question is why. Why is his WARP low? Low amount of steals and inefficiency in finishing at rim. Okay, is that correctable? How much of that is coaching style? How much of that is mental growth? How much of that is physical growth? How much of that is skill that could be developed? How much will that change from rule changes?

Steals are a perfect example. Lots of these statistical models place a huge emphasis on steals. There's some validity to it. But Bill Self's defensive system is very much against gambling. Since Bill Self took over the Kansas program in 2004, Kansas as a team has been in the top 100 in steal % once. Once. And they were barely in the top 100, at 86th in 2011. This is a team that has gone 325-69 over that span. The question then becomes, what could Wiggins have averaged if he played on a different coach, with a different scheme?

(Let's also ignore the fact about how steals are awarded by scorekeepers).

As Rich said, I think ball handling is Wiggins swing skill. If he tightens that up, and gets more confidence in executing hesitation and misdirection moves, he'll get significantly better looks in the half court, and settle for pull-up jumpers less. The question isn't so much "did his lack of ball handling hinder him", because we know it did, but "can he improve his ball handling in the future." We're talking about a kid who played at 18 years old for most of the year.

What I will say is that if Andrew Wiggins doesn’t improve upon his ball handling and/or ability to finish at the rim, he will be disappointing in relation to where he’s drafted. Still a starter, based on transition, solid catch and shoot, and defense, but nothing special. And I think the stats show that. If he improves his handle? I think he can be an all-star. Improves handle and improves finishing ability? Better than an all-star (although not LeBron/KD level. I think Paul George is a solid upper end).

More or less, I think these models get more accurate the further a player is along in his development. It's partly why most of the "outliers" have been athletic freaks who had yet to develop their skills to the point where they could fully take advantage of ithem.

Another example is, I'll look at the stats, and it'll show me "wow, this guy's efficiency drops big time when he drives to his left." Why is that? Does he settle for jump shots more? Is his handle with his left hand weak?  Does he struggle to finish at the rim with his left hand? Which of these can be corrected? Even situational data as specific as that needs context.

"Why, what, and what will be." I think these models do a fairly good job of one of those 3. Traditional scouting will never be gone from the equation because stats will always need context. Using them without putting them into context, regardless of how well founded the statistical model is, will always yield "outliers".

As Hinkie said when he was hired, it's just one piece of the puzzle.