Today, I have a fun experiment that I’d like to share with all of our loving readers. In preparation for the draft, I started doing some digging into the 2014 NBA draft class. What I hoped to accomplish with my analysis was a prediction of how good players from the upcoming draft class could be, based on their college statistics. There are factors that are out of our control when trying to predict how college athletes will perform at the next level (injuries and such), but we can certainly hope that there is a correlation between a player’s performance in college and their effectiveness in the NBA.

Let me start off by explaining how I decided to perform this analysis. I decided to perform a linear regression of players that are currently in the NBA from the previous 3 draft classes, based on their final college season. While I did not choose a random assortment of players, I tried to select a variety of different positions and performance levels for each of the players. The list of players that was selected for this analysis includes (in no particular order):

2011 Class:

2012 Class:

2013 Class:

2011 Class:

- Kenneth Faried
- Kyrie Irving
- Klay Thompson
- Kawhi Leonard
- Tristan Thompson
- Derrick Williams
- Brandon Knight
- Chris Singleton
- Kemba Walker
- Marcus Morris
- Nikola Vucevic
- Norris Cole

2012 Class:

- Bradley Beal
- Anthony Davis
- Andre Drummond
- Damian Lillard
- Dion Waiters
- Andrew Nicholson
- Austin Rivers
- Harrison Barnes
- Jared Sullinger
- Kendall Marshall

2013 Class:

- Anthony Bennett
- Michael Carter Williams
- Cody Zeller
- Ben McLemore
- Kelly Olynyk

*Note: I chose players who had reasonably consistent playing time (outside of Bennett) in order to ensure that there was enough data to support the type of production they were creating in the NBA. Hence, the 2013 class, which was one of the poorest in a long time, had very few individuals that seemed to qualify. Furthermore, the number of subjects was limited to reduce the strain of manual data entry, which sucks.*Now, the stats that I chose to look at for this study (from their last college season) were Simple Rating Score of opponents (SRS, a measure of the competition level for each game), minutes played (MP), effective field goal percentage (eFG%), true shooting percentage (TS%), offensive (ORB) and defensive (DRB) rebounds, assists (AST), steals (STL), blocks (BLK), turnovers (TOV), and points (PTS).

By using both effective and true shooting percentages, I hoped to eliminate some of the biases there may be between bigs (lack of 3 point shooting) and wing players (fewer under the basket shots). Furthermore, the SRS was included in an attempt to eliminate the bias of putting up big numbers against lower level competition, which could heavily influence players in small schools. Then I chose Player Efficiency Rating (PER) as the statistic to provide a comparison for what their NBA production at the next level would be, being the statistic that is least influenced by the team the player has around them.

The result of the regression came up with the following equation to determine a current college player’s NBA PER:

PER = 0.52 (SRS) - .06 (MP) – 7.16 (eFG%) + 11.36 (TS%) + 1.20 (ORB) + .22 (DRB) + .91 (AST) + 1.06 (STL) + 2.41 (BLK) – 1.14 (TOV) + .16 (PTS)

When we apply the equation to the players used for the regression, we see the following results:

By using both effective and true shooting percentages, I hoped to eliminate some of the biases there may be between bigs (lack of 3 point shooting) and wing players (fewer under the basket shots). Furthermore, the SRS was included in an attempt to eliminate the bias of putting up big numbers against lower level competition, which could heavily influence players in small schools. Then I chose Player Efficiency Rating (PER) as the statistic to provide a comparison for what their NBA production at the next level would be, being the statistic that is least influenced by the team the player has around them.

The result of the regression came up with the following equation to determine a current college player’s NBA PER:

PER = 0.52 (SRS) - .06 (MP) – 7.16 (eFG%) + 11.36 (TS%) + 1.20 (ORB) + .22 (DRB) + .91 (AST) + 1.06 (STL) + 2.41 (BLK) – 1.14 (TOV) + .16 (PTS)

*Note: All coefficients are rounded and the statistics that are used for the college player are their average of those statistics for their final college season.*When we apply the equation to the players used for the regression, we see the following results:

There are two shocking results that I can see early on with this data. First, Anthony Davis is an absolute beast, and has been the best player to come out of the past three drafts (not even close). Second, the analysis tends to favor bigs (hence, Thompson), which is probably a result of favorable measures for rebounding and shooting percentages. Make sure to keep these factors in mind when we look into the data for the upcoming draft class. (Third note that probably only I care about: Bennett wasn't dead last!).

Now, for the players that were chosen for the upcoming class, I tried to grab as many players as I could within the predicted top-10 spots in the draft. I could not unfortunately evaluate Dante Exum, or other international prospects, as the international scene is very difficult to predict how well a player’s statistics will translate, and varies largely on the league that the player is playing in. The following players were evaluated:

Using the same process as above, the results were as follows:

Now, for the players that were chosen for the upcoming class, I tried to grab as many players as I could within the predicted top-10 spots in the draft. I could not unfortunately evaluate Dante Exum, or other international prospects, as the international scene is very difficult to predict how well a player’s statistics will translate, and varies largely on the league that the player is playing in. The following players were evaluated:

- Joel Embiid
- Tyler Ennis
- Aaron Gordon
- Gary Harris
- James McAdoo
- Doug McDermott
- Mitch McGary
- Shabazz Napier
- Jabari Parker
- Julius Randle
- Marcus Smart
- Nik Stauskas
- Noah Vonleh
- T. J. Warren
- Andrew Wiggins

Using the same process as above, the results were as follows:

While this is simply a prediction, there are a few points that I want to make about this that may help shed some light on what I see with these results:

- If these results hold true, this draft class, while clearly better than average from the previous 3, does not have a true superstar in the mix.
- Embiid may have came out at the top of the predicted PER, but that does not mean that he will produce from the get-go. Bigs generally have a difficult time adjusting to the jump from college to the NBA as a result of bigger and more skilled defenders. Health is also a major factor here, and a justifiable reason why Embiid may end up dropping in the draft.
- Parker, Wiggins, and Randle all came out slightly above average, but still did not come out as the high level players that they are predicted to be at the next level. Wiggins is the only one who might have a strong argument to look away from this prediction though, based on the general play he had in college, in which he showed absolute dominance when he really wanted to, yet tended to shy away at times.
- Moondog Landing favorites Nik Stauskas and Doug McDermott came out on the bottom… Lets hope our eyes are a little better than this analysis.

I hope that you have found this analysis to be insightful and helpful in your upcoming draft studies. Please feel free to leave some feedback on the process that was used, or even your thoughts on the prospects that were studied here.

-- BA

-- BA