Please, Stop Yelling at Neil Paine for his Andrew Wiggins Article

facebooktwitterreddit

Jan 17, 2015; Denver, CO, USA; Minnesota Timberwolves forward Andrew Wiggins (22) shoots a free throw during the second half against the Denver Nuggets at Pepsi Center. The Timberwolves won 113-105. Mandatory Credit: Chris Humphreys-USA TODAY Sports

We all love to criticize talking heads and pundits for archaic anti-numbers points of view. Heck, over at the Sports Analytics Blog, we’ve had our fair share of fun with folks like Harold Reynolds and Steve Dilbeck. This is a thing that happens and it happens often.

It’s often unfair. We love to criticize and poke fun, but we don’t spend as much energy as much to defend and add nuance. All this leads to the recent controversy surrounding FiveThirtyEight’s Neil Paine and his much-discussed article on Andrew Wiggins. This time, it’s the opposite, where a numbers-focused writer stirred up the masses. The headline, Paine has admitted, left a lot to be desired. But we all keep talking. And it deserves a second look at the issues at hand.


“Vashro’s probabilities give Wiggins a higher shot at being a star and at not being a bust, but he’s got much higher odds of just being a bench player.”

I’ve struggled with the meaning of that sentence for a long while. It’s a sentence that I wrote back in June, comparing Andrew Wiggins and Jabari Parker. The sentence was in a mega-post over at WatingForNextYear that I titled An Analytics Reader’s Guide to the 2014 NBA Draft. The sentence referred to the draft projection chart created by Layne Vashro, a snippet of which is shared below.

Eyeball those projections for Parker and Wiggins. For the Duke forward: 8 percent bust, 16 percent bench, 59 percent starter and 18 percent star. For the Kansas guard: 2 percent bust, 48 percent bench, 24 percent starter and 26 percent star. So yes, from Vashro’s numbers, Wiggins somehow managed to have higher shots at being a star and not being a bust. His odds of being “just” a bench contributor were also three times higher.

This is what we talk about when we talk about the statistical background of Andrew Wiggins as a lightning-rod analytics topic. His numbers at Kansas left a lot to be desired. Jacob Frankel described his collegiate production as “Not bad, by any means, but just nothing impressive.” You could point out the mediocre 49.9 effective field goal percentage. You could point out only 1.5 assists against 2.3 turnovers per game. You could point out lots of things [1. One such thing being his 26.4 percent usage rate, which was fairly high, but obviously not really close to Jabari Parker’s 32.7 percent usage (Requisite link to expected usage-efficiency curves via now-Houston Rockets VP Eli Witus).]

That is what led to many fairly aggressive analytics-related takes on the Wiggins debate even before he played an NBA game. I’m not intending to give credence to some of these, but just hope to share that they were out there. Taking Wiggins No. 1 “would be among the most egregious draft blunders of all time,” said Dean Demakis. “Spending a lottery pick on him is downright irresponsible,” said Arturo Galletti. The “most likely outcome for Wiggins at this point looks more like quality role player than star,” said ESPN’s Kevin Pelton.

So the book was out on analytics models not caring too much for the 19-year-old Canadian. Of course, the best models also include qualitative scouting information too. And, yes, those models included the fact that age is so important when it comes to projection models, as Andrew Johnson wrote back in May. But for all of the talk about potential, Jabari Parker actually is three weeks younger. What made us say one had more possible long-term improvement than the other[2. I’m also in love with Layne Vashro’s article on the topic of potential and how we measure it. Does Wiggins’ elite athleticism give him an added potential advantage? We don’t know concretely.]? The data challenged conventional wisdom. There’s value to that.


Fast forward 26 games into the NBA season. The Minnesota Timberwolves, who were struggling mightily without Ricky Rubio, were 5-21. Wiggins had started every game and averaged 30.7 minutes per contest. His shooting efficiency numbers, more so than at Kansas, were quite dreadful: 40.5 effective field goal percentage. He still only had 1.3 assists against 2.0 turnovers. His 21.4 percent usage rate was only barely above an average player.

This is the basis that led to Neil Paine’s revised article on Wiggins’ long-term future, after his 27th professional game. Here is a player, whose statistical profile already was critiqued in the 35-game college sample. And now his NBA efficiency was far worse with lower offensive involvement.

The transition to the NBA is obviously never seamless, but this was a disappointing start, no doubt. And that is the point that Paine made in the actual content of his article, if you read past the headline.

Paine calculated Wiggins’ Statistical Plus-Minus[3. Yes, of course, there are caveats with any one-number-fits-all type analysis, as Seth Partnow mentioned preseason. But let’s continue.] to be -4.8, worst in the NBA at the time. But factoring in the standard error for that number, his draft position’s usual SPM and the standard error for that, Paine wrote that the best guess for Wiggins’ season was really a -1.9 SPM. So he combined the two together – the actual results in 26 games and expected production for the rest of the season – and calculated a -2.9 SPM for his projected rookie year[4. Note: These are point estimates. There are confidence intervals for these numbers. They’re not exactly spot on; they’re just trying to find a true single number on the results we’ve seen so far. There is a large range, still.].

To put everyone on equal age footing Paine then calculated an age-22 equivalent +0.2 SPM for Wiggins. That helped his case due to his young age compared to most NBA rookies historically. But the statistical evidence was mounting against him ever having a superstar-like peak. The age-22 equivalent SPM was closer to names like Josh Childress, Ben Gordon, James Posey [5. Paine has somewhat defended the Posey comparison, albeit critiquing the headline. Posey had a very solid career in many different roles, averaging 27.5 minutes per game in 74 contests in his first 11 seasons. He was perhaps the prototype for an above average, yet un-star-like rotation player.] and Robert Horry. This was Paine’s point. A snippet of his chart is shared below.

If you want to point out that SPM maybe lacks some nuanced defensive data where Wiggins might excel, Nylon Calculus editor Ian Levy also debunked that idea at Vantage Sports. Levy pointed out that his advanced defensive metrics were fairly mediocre, at best, and nowhere near another rookie in this class, K.J. McDaniels. There wasn’t enough data for aging curves and projections, but the stats there weren’t encouraging there either.

All of that was true in 26 games, about one-third of an NBA season. Wiggins was not performing well. His college stats – which Paine did not factor in – weren’t encouraging. In a holistic evaluation process, you simply can’t ignore this data. You can’t only rely on qualitative reports. Data picks up on so many things that the naked eye doesn’t. That was what happened when Neil Paine wrote his article on Dec. 24.


Now, of course, Wiggins is on fire. Starting with the highly touted matchup against the Cleveland Cavaliers, he’s been wowing the NBA world left and right. In 14 games and 37.1 minutes per contest, he is averaging 20.7 points, 4.7 rebounds, 2.6 assists, 1.9 turnovers and 1.2 steals. The shooting efficiency is up to an above average 52.7 effective field goal percentage mark. He’s doing that on increased involvement with a 23.5 percent usage rate. It has been a very impressive month.

But Wiggins’ improved play over the last month doesn’t disprove anything that Neil Paine wrote on Dec. 24. It doesn’t mean that the statistical evidence of Wiggins’ 35 college games and first 26 NBA games didn’t show reason for concern. Obviously, his rest-of-season projection will now be incredibly better. But that happens. Uncertainty happens in all of professional sports. Decisions can’t be made solely on scouting reports nor solely on the data. Paine[6. On Paine: He’s been one of the leading researchers, innovators and supporters of basketball analytics for many years. He’s an incredibly bright guy, one who used to work for Basketball-Reference and the Atlanta Hawks. He’s over-qualified to be “just” a sports analytics writer at FiveThirtyEight.] did what he could with the numbers at hand and his very solid analysis was clouded by emotional responses to a headline.