NBA Draft: Why College Points Don't Matter but Assists and Rebounds Do

In both college basketball and the NBA, points are the most glamorous of all stats. People recite the scoring averages of stars far more often than the rebounding or assist averages. SportsCenter highlights show three-pointers and rim-rattling dunks much more often then tough boards or beautiful, yet simple passes.

Why? For the very same reason that we go shoot hoops in the driveway and practice our long-range bombs instead of having someone else shoot so we can grab rebounds time after time.

Scoring is fun. There simply isn't any doubt about that.

But for NBA prospects, scoring at the college level is actually almost irrelevant. I say almost because obviously it's important that a prospect has the talent to put the ball in the basket, but almost every single one of them does. The specific numbers are the things that don't matter. 

It's often said that rebounding is the stat that translates best from the college level to the professional game, but I've never seen any factual evidence to back that claim up so I set out to determine its validity myself. 

Looking back at the five most recent drafts, I analyzed the players' rookie per-game averages in the three categories (points, rebounds and assists) as well as those same averages from each player's final year spent at the collegiate level.

The reason I only focused on rookie seasons was because I wanted to see the actual transition from college to pros and not the development that occurred in the ensuing campaigns under NBA tutelage. 

Then I thought to myself that not all rookies play a lot, so perhaps it would be better to analyze their per-36 minutes numbers. Sure enough, there was a dramatically better correlation between per-36 averages in college and as a rookie than there was between the straight-up per-game averages.

I also only looked at first-round draft picks who played at least 10 minutes per game to counteract the confounding effect of lower-end talent and small sample size. 

Sure enough, as you can see in the above graph (college points per 36 minutes are on the X-axis and rookie points per 36 minutes are on the Y-axis), there isn't really any correlation between the two numbers. The r^2 correlation coefficient of 0.192 indicates that there's a slightly positive trend for the best-fit line, but college points are in no way predictive of NBA points. 

If a player can score in college, he will most likely be able to score at the NBA level, given enough playing time. Likewise, if he has trouble finding the bottom of the net at one level, that difficulty will transfer to the next level with him.

But if one player scores 24 points per game in college while another scores just 18 per contest (assuming they play the same number of minutes), it does not necessarily follow that the first player will score at a higher rate than the second in the Association. 

So how about rebounding?

As you can see from the above graph, set up in the same way as the previous one for points, there is a remarkably stronger correlation. In fact, seeing as there is a correlation coefficient of 0.67 (above the generally accepted 0.6 minimum threshold in the scientific community), I feel confident in the ability of the best-fit line to predict a player's rookie per-36 rebounds per game given the corresponding collegiate mark.

Most players will vary a bit from the results of the best-fit line, but as a whole, there is clearly a trend.

The same thing applies to dishing out the dimes, but this time to an even greater extent. Kind of.  

In the above graph there is a rather large cluster of players near the origin, most of which are the frontcourt players whose assist numbers are rather depressed because of their role within the offense. When you include those frontcourt players, r^2 for the best-fit line is 0.706, the highest of all three stats. 

But, if you take out all non-guards, r^2 drops to 0.612. Therefore, collegiate assist numbers can still be predictive of their NBA counterparts, just not to the same extent as rebounding numbers. 

All in all, both the claim whose validity I doubted and the claim I made at the beginning of this article are true. Rebounding is indeed the stat that translates best from one level to the other and scoring at the collegiate level doesn't really matter all that much. 

So next time you sit down to watch a future NBA player do his thing, try not to focus on the results of his shots. It's hard to do, but it's the right way to look at the prospects.