The difficulty of evaluating wing prospects for the WNBA draft with statistics

Over the past few years that I've been looking closely at the statistics of WNBA prospects, some pretty basic position-by-position principles are have been reinforced as more prospects are added to the pool of history.

Sure there are exceptions to any rule based on intangibles or random quirks, but you can generally feel pretty secure in separating a player who will be worthy of a roster spot from one who probably won't make it (at least without some seasoning) in the post and at point guard. I've figured out (roughly) some indicators for mid-major players. And the experimentation with similarity scores has been extremely helpful.

But one type of prospect has actually gone the exact opposite direction, getting to a point where I was almost ready to scrap everything I thought I had figured out: non-point guard perimeter players. As in, all wings. And it actually got worse when I started adding the full set of drafted and undrafted wings from the 2008 draft since the 2013 draft.

As an example, let's compare two anonymous prospects (who will be revealed at the end of this piece) that seemed to leap off the screen as particularly confounding as I looked over the raw data:

Prospect	Height	PPG	2pt%	3pt%	FTP	Ast/40	Stl/40
Player A	5'8"	19.9	49.44%	32.2%	14.50	3.93	1.82
Player B	5'7"	16.6	41.73%	36.8%	16.52	3.70	3.41

Senior year numbers for two past guard prospects.

Prospect	Height	PPG	2pt%	3pt%	FTP	Ast/40	Stl/40
Player C	6'0"	15.0	46.19	35.85	17.61	4.67	2.88
Player D	6'0"	10.8	45.62	33.71	15.36	3.01	1.30

Senior year numbers for two past small forward prospects.

All of four of these prospects played in major conferences and all but one helped their team to the NCAA Tournament. And every one of these players has significant flaws that could lead to concerns about whether they'd make it in the league:

• Both Player A and Player D share a problem that can suggest that they're not an elite athlete: the combination of a low free throw production rate (FTM/FGA) and low steal rate. One might expect a wing's athleticism to show up offensively in their ability to drive to the rim and draw contact or defensively in their ability to force steals. With the former in particular, a low free throw production rate usually reflects a player settling for a lot of jumpers rather than aggressively going to the basket.
  
  For Player A, that's a problem compounded by her height. For Player D, the problem is compounded by having a mediocre 2-point percentage for a wing prospect - nobody wants a wing who lives on jump shots but won't be able to make them often.
  

• Player C also has that free throw production concern, but the combination of steals and assists bodes well for someone who is a) versatile and b) capable of making plays on both ends of the floor. In addition, her 2-point percentage is passable at 46%.

• Player B is a player whose ability to become a long-term contributor would be questionable. As an inefficient scorer at 5'7" - and among the purest scorers of any prospect in the last six drafts - it looks like she'd have to play point guard; her relatively low assist rate and the fact that she didn't really play point guard in college cast a shadow of doubt over whether she'd be able to do that. One thing should keep her on the radar though: her steal rate ranks 8th among major conference wing prospects in the last six drafts, which reflects elite athleticism. And inefficiency aside, she obviously knew how to score: she led her team in scoring twice in her four years and the entire conference in her junior year.
  

Strictly based on the numbers and what statistical data we have about the importance of 2-point percentage and steal rates for NBA prospects, I might suspect that all of these players would get drafted in our hypothetical draft, but that Players B & D would struggle to stick anywhere while the other two would be solid contributors but not quite stars. Here's how things actually turned out:

• One player was drafted in the first round and three of these players were drafted in the second round.
• One never made a roster, another only lasted a year, one ended up making the All-Rookie team and earned another postseason award during their career, and one became an All-Star.

The statistics listed above are the box score-based statistics that I've usually thought of as important when looking at prospects, but clearly whatever rudimentary principles I thought I had working for me failed here.

So while the math-phobes among us will simply say "SEE! Numbers and, by extension, MATH suck! Can't truss it!", there's probably a middle ground that can be framed as a question: what numbers or context are we missing?

And that's probably the best way to think about draft prospect statistics: what key questions do they illuminate? And what contextual factors - the situation that produced those numbers - are left out?

Clearly, we just can't quantify human potential, work ethic, or personality variables, but I'll present a few missing elements that might have helped to better evaluate these players on paper.

What type of data is missing from our discussion of WNBA draft prospects?

• How much would a WNBA draft combine help? The annual NBA draft combine is immensely helpful for basketball junkies in evaluating prospects for two reasons, among others: you get actual heights and wingspans. A lot of the other stuff can get fluffy and misleading, but because college programs often exaggerate heights by an inch or two (which is a big deal for post players, especially) and we don't otherwise have wingspan data, bringing back the combine could be useful.
  
  And I will say that it's highly likely that players B & D would have seen their stock go up a bit with combine data of the scale available for NBA prospects. Of course, many a NBA GM has been duped by combine All-Stars, who are great at showing off when they're by themselves in a gym without a basketball but considerably less adept at impacting actual basketball games.
  

• How much does size matter? Even with combine-like data, it's hard to know how much height even matters for wings based on the data we do have. An 6-foot guard lacking athleticism is going to struggle at the next level no matter how good their numbers are; players with suspect statistics have overcome that due in part to being athletic enough to get things done against pros (Player B being one example). Then there are a few college point guards who have made it in the league as shooting guards. Of note though is that the shortest wing prospect (using the WNBA.com prospect data) in the last six drafts has also been one of the best in their class: Player B.
  

• How important is mid-range shooting ability? The prevailing wisdom in basketball these days is that 3-pointers, shots at the rim, and free throws are the most efficient way to score - it's difficult to dispute any of that. But we've also seen players who thrive on mid-range shooting - again, Player B being one of them - end up making rosters. What would be interesting to learn more about is how much shot location data matters: is there a ratio of shots at the rim, jumpers, and threes that end up making or breaking a prospect?

• How much does the system a player played in during college matter? The ability to score unassisted baskets at the rim especially is probably a good indicator for a wing prospect. Usage percentage helps get at that, but all four of these prospects were relatively high usage players for their team and the results were all over the place. But the number and quality of unassisted scoring opportunities a player gets is probably directly related to the type of offense their team runs (and perhaps their pace of play). Related to that, if a coach recruits specifically for a given style of play - for example, a coach who runs a methodical offense recruiting good catch and shoot and passing players - might it be reasonable to make assumptions about what they're capable of in terms of getting unassisted attempts?

• What to do about defense? It's really hard to know what to say about defense: obviously disinterested defenders have made rosters and thrived; conference Defensive Player of the Year award winners have entered the pros and have almost looked lost defensively; players known for their scoring ability have come into the pros and earned their minutes defensively. Even if we could quantify defensive potential - and we really can't, particularly given strength of schedule issues at the college level - the fact remains that effort and a player's fit in a coach's defensive scheme will end up trumping most of what we could say, especially on the wing.
  
  Nevertheless, it has to be mentioned that Player D won her conference's Defensive Player of the Year award twice and has been considered a solid defensive player in the WNBA, though she hasn't won any defensive awards. Player B, perhaps unsurprisingly based on her college numbers, ended up ranking fourth in the WNBA in steal percentage in her rookie year; steals aren't necessarily a reflection of "defense", but there is something to the notion that steals reflect a skill that transfers well from college to pro.
  

So those are a lot of unknowns, or at least questions to answer, about these prospects. And as you can probably tell by the players mentioned above, those questions have an impact on how we evaluate the anonymous prospects above: Player B and D both stand out as players who probably offered something that we can't pick up on simply by looking at statistics.

And as it turns out, they were definitely the best prospects of this bunch.

Prospect	Height	PPG	2pt%	3p%	FTP	Ast/40	Stl/40
Jolene Anderson, Wisconsin (2008)	5'8"	19.9	49.44%	32.2%	14.50	3.93	1.82
Riquna Williams, Miami (2012)	5'7"	16.6	41.73%	36.8%	16.52	3.70	3.41

Prospect	Height	PPG	2pt%	3pt%	FTP	Ast/40	Stl/40
Tanisha Smith, Texas A&M (2010)	6'0"	15.0	46.19	35.85	17.61	4.67	2.88
Essence Carson, Rutgers (2008)	6'0"	10.8	45.62	33.71	15.36	3.01	1.30

The whole reason I went through this exercise is because Carson and Williams really stand out as perplexing cases statistically, which led to a question: is there any purpose in using statistics to evaluate prospects?

After forcing me to take a closer look at things and consider some different approaches to wings based on what others have done with NBA prospects I concluded that the answer is "yes" - Carson, Smith (in not making a roster) and Williams just stand out as possible exceptions that don't disprove any rules.

Smith reportedly told the Seattle Storm she wanted to prioritize finishing school over fighting for a roster spot in 2010, which means her story is sort of incomplete. Carson was not really a bad prospect statistically, but there are a number of players with better profiles who didn't make it - Smith included - which makes her an odd case (or just another sign that it's easier to figure out who might make a roster than it is how good a contributor they'll become).

The other two are much trickier.

Anderson had her flaws - the combination of a low free throw rate and below average steal rate at 5'8" is often enough reason to question a perimeter player's ability to compete in the pros - but overall, those numbers look much better than those of Williams given the performance of past prospects.

Looking at the numbers alone, Williams simply wasn't that strong a prospect - comparing her statistical profile as a 5'7" player to other past prospects, she almost looked like a non-prospect. Yet as much as Williams is something of an anomaly statistically, she might also be the clearest example in recent memory where there was some obvious dissonance between the statistics and observation that ended up being resolved by the power of observation.

As her college stats suggested, she has hardly been an efficient scorer as a pro. And to reinforce the point that she's an anomaly, part of Williams' success is indisputably related to her fit with the team that drafted her: the Shock have been one of the fastest-paced teams in the league since the relocation to Tulsa, which is a great fit for Williams.

But the key is that Williams is an elite athlete even among the elite class of athletes that ends up making it in pro basketball. I'd hesitate to say the best athlete in the last six years, but all you had to do to temper the statistical story was watch her play a couple of games: she could get almost wherever she wanted on the court and whatever shot she wanted. Even if her shot selection - or off the court controversy - didn't suit your tastes, it was impossible to deny that she had at least some small shot of competing with the best if given the opportunity.

However, as you eyeball all the data it's also clear that Williams is an exception that doesn't really disprove some of the larger trends: athleticism is an important attribute in a league with limited roster spots and her steal rate does suggest she has that in spades. And it's still true that most players with 2-point percentages below 43% just don't make it. As more than one person suspected, Williams' athleticism has been a perfect fit for Tulsa's uptempo style of play and it's not exactly surprising that she ended up succeeding in spite of suspect statistics.