Last year I moved from a team-level predictions model to a player-based predictions model. It was a large undertaking, and there was one big piece of information that I was not able to incorporate in the model in the first year: recruiting star ratings.

In the team-level model I had historically only focused on the Top 100 recruits. That was because players ranked outside the Top 50 rarely impact their team during their freshman season. But this summer I have been working with the fine folks at verbalcommits.com to look more closely at the recruiting star ratings for every D1 player. Verbal Commits has compiled the star rating recruiting information from ESPN, Rivals and Scout for the last six years. (And because some of those players were seniors six years ago, I actually have star ratings for a limited number of players going back nine seasons.) I have combined VC’s recruiting data with the college stats of those players. And I now have 19,000 individual seasons of star ratings and player data.

With this combined data I have been able to determine that star ratings have a lot of predictive power that lasts far beyond a player’s freshman season. This first table shows how the average Points Per Game production varies based on a player’s star rating and class:

Points Per Game

Fr

So

Jr

Sr

2 Stars

3.5

5.0

6.1

7.2

2 to 3 Stars

4.1

6.2

7.7

8.3

3 Stars

4.3

6.7

8.1

9.1

3 to 4 Stars

4.6

7.8

8.7

10.1

4 Stars

5.9

8.3

10.0

10.7

Over 4 Stars

7.5

11.2

12.8

12.1*

Note: Verbal Commits truncates the star ratings at 2 stars because ESPN, Rivals and Scout do not all give out ratings below two.

*Don’t overlook the fact that attrition matters. Elite productive players are more likely to leave for the NBA which explains the slight drop in PPG production in the “Over 4 Stars” group.

Regardless, the average two-star senior produces only 7.2 points per game. As a freshman, the average player rated over four stars scores 7.5 points per game. Clearly, it is good to be an elite recruit. Now often PPG production can be misleading. We need to look at efficiency (ORtg), minutes, and a player’s aggressiveness on the court (the percentage of possessions used) to understand what leads a player to score at a high rate. But it turns out that all three of these factors matter, and that the Points Per Game production gives us a fairly accurate picture here.

The next table shows the average percentage of minutes based on star rating and a player’s class:

Pct Min

Fr

So

Jr

Sr

2 Stars

25%

34%

40%

46%

2 to 3 Stars

30%

42%

49%

52%

3 Stars

30%

44%

51%

57%

3 to 4 Stars

33%

51%

54%

59%

4 Stars

39%

51%

57%

63%

Over 4 Stars

47%

63%

67%

67% 

These tables are the average for all D1 schools. If you focus on power conferences, the trends are the same, but players with lower star ratings play even fewer minutes. Conversely, at the smallest conferences, two- and three-star players play more.

A player’s average aggressiveness shows a similar pattern:

Pct Poss

Fr

So

Jr

Sr

2 Stars

18%

18%

19%

19%

2 to 3 Stars

18%

19%

20%

20%

3 Stars

18%

19%

20%

20%

3 to 4 Stars

18%

19%

20%

21%

4 Stars

20%

21%

21%

22%

Over 4 Stars

21%

22%

23%

21%

And ORtg shows the same pattern as well:

ORtg

Fr

So

Jr

Sr

2 Stars

91.7

96.3

97.8

100.6

2 to 3 Stars

93.8

98.6

101.2

102.0

3 Stars

95.3

100.4

103.5

104.9

3 to 4 Stars

96.8

103.2

105.3

106.2

4 Stars

97.8

102.8

105.3

107.5

Over 4 Stars

103.0

108.8

110.0

109.6

While these raw averages contain a lot of information, while working with the Verbal Commits data I have also discovered two other key facts:

1) Star ratings continue to have meaningful predictive power even after we know about a player’s previous college stats.

For example, in the above table the average two-star sophomore has an efficiency of 96.3, while the average three-star sophomore has an efficiency of 100.4. That’s a difference of just over four points per 100 possessions. You might think these differences were all about the different skill levels when these players enrolled in college. But that isn’t the whole story. Even if a two-star and two-star player have equivalent freshman seasons (for example, if the two players both have an ORtg of 95 during their freshman season), on average the two-star sophomore and the three-star sophomore will have ORtgs that differ by 2.5 during their sophomore season. Even after accounting for college stats, the player with the higher recruiting rank will come out ahead, on average.

The reason is that high school star ratings are not just about how polished a player is headed into college. Star ratings also incorporate some information about a player’s ceiling. Even if they have the same freshman season, statistics confirm that the three-star sophomore has higher potential.

2) Other recruiting information has predictive power too, not just the star rating.

The recruiting services do a solid job evaluating three-, four-, and five-star players. There will always be noise and inconsistencies, but by attending the AAU and elite high school events, the scouts get multiple opportunities to evaluate the top players.

But evaluating two-star and lower prospects is notoriously impossible. They don’t get invited to the same elite events. And when they do play with other elite players, they often get overshadowed. Evaluating these players given all the major differences in high school competition is very hard.

Nonetheless, acting on a hypothesis from Paul Pettengill of verbal commits, I can tell you that there is additional information which predicts the performance of lower ranked players.

It turns out that the number of offers a two-star player receives has some predictive power for his college stats. Similarly, if a player receives an offer from a higher ranked program, his performance will also be higher. To summarize, an Iona enrollee that had an offer from Temple, Seton Hall, Fordham, and American will perform better than a player whose only offer was Iona, even if they were both evaluated as two-star prospects.

Bottom Line: To many reader’s today’s column may seem obvious and boring. Everyone knows that getting more highly ranked recruits will lead to more success. But not every piece of conventional wisdom is confirmed by the data. Sometimes we see enough players fail to develop and wonder if the recruiting rankings really matter. And while a players potential is far from the only thing that matters, (teammates and coaches can be even more important), there is no question that a players work ethic and athleticism is on display at the high school level. Recruiting rankings matter, and not just for the Top 100.