FanPost

Effect of Recruiting Class Ranking on Future Success: A Statistical Analysis

Since Maryland has the 8th best recruiting class according to ESPN, I was interested in seeing how that historically has related to future success. I was wasting the day away at work and ended up doing a very, very rough and probably not very statistically valid analysis. I set up a multiple regression model to examine the effects of Incoming Recruiting Class Ranking (IRCR) and Previous Year End of Season Rank (PYESR) on the coming End of Season Rank (ESR). I figured this should be an alright and reasonable place to start, even though there are absolutely other variables at play (I think percentage of players returning, or number of seniors or something like that, maybe percentage of returning scoring probably factor in as well, but I didn't know how to easily find that data, so we'll have to live with it). Let me say before I continue, that the results are by no means conclusive or even altogether meaningful, but at least it's an opportunity to have an interesting discussion about what other variables really do go into End of Season Rank and how we think Maryland will fair this coming year.

To measure ESR and PYESR, I used KenPom's final standings for each year. To measure RCR, I used ESPN's yearly recruiting class rankings. I actually only took the top 15 teams for each year (partly because it was slightly less work, and partly because Maryland is in the top 15 so I thought that was good enough). I used 6 years of data (2008-2009 season to 2013-2014 season), so altogether I had 90 data points.

First, some summary statistics. Out of the 90 observations (all teams in the top 15 for recruiting classes from each of the past 6 years), 47 finished in the top 30 in Kenpom the following season, or 52%. There were 16 teams that had a top 15 recruiting class and also finished between 30 and 50 in Kenpom the previous season, and 4 of these teams (25%) finished in the top 30 in Kenpom the following season.

Now for the more advanced stuff. I ran a multiple regression in Excel and got the following output:

SUMMARY OUTPUT

Regression Statistics

Multiple R

0.474540396

R Square

0.225188587

Adjusted R Square

0.207376831

Standard Error

42.10786391

Observations

90

ANOVA

df

SS

MS

F

Significance F

Regression

2

44832.81837

22416.40918

12.64269393

1.51365E-05

Residual

87

154257.2816

1773.072203

Total

89

199090.1

Coefficients

Standard Error

t Stat

P-value

Intercept

11.90391355

9.676603071

1.230174831

0.221947664

IRCR

1.578681489

1.051681457

1.501102333

0.136949818

PYESR

0.576804716

0.132086907

4.366857607

3.46319E-05

Basically, the model as a whole is statistically significant as is. The two independent variables I chose only explain about 22% of the variation in the dependent variable, which is sort of low but still ok, indicating there are other factors affecting End of Year Ranking (which we knew).

To explain the coefficients, according to this model, for every extra increase in Class Rank (from being Ranked 5 to being Ranked 4, for example), you can expect your End of the Season Ranking to get better by about 1.57. And for every extra increase in Previous Year End of Season Ranking (from finishing 32 in Kenpom instead of 33), you can expect your End of Season Ranking to get better by about .57. The caveat here is that the coefficients for the intercept and for Incoming Recruiting Class Rank are not statistically significant on their own, though the coefficient for IRCR is close. Even with the understanding that the ICR coefficient isn't statistically significant, the model suggests that ICR has approximately 3 times the effect on ESR as does PYESR.

Ok, so we have some working model. It isn't necessarily a reliably predictive model, but we can now use it to get the expected End of Season Rank for teams for next year based off of their recruiting class ranking and this past year's finish. I applied these coefficients from the model to get the expected 2014-2015 finish for teams in the top 40 Recruiting Class Rankings this year using the regression equation:

ESR = 11.9 + (1.57*IRCR) + (.57*PYESR)

Maryland, with a recruiting class ranking of 8 and a previous year finish of 40 in Kenpom is expected to have an End of Season Kenpom ranking of 47. Before you think we're going to be worse than this past year, notice that this number is misleading, since in this model, the best you can possibly do (if you had the best ranked recruiting class and also finished #1 in Kenpom last year) is 14. So I ranked in order of expected finish all of the teams in the Top 40 of this year's recruiting class rankings:

Team

Incoming '14 Class Rank

2014 Finish

Expected '15 Finish

Louisville

4

1

18.79544422

Duke

1

13

20.98105635

Kentucky

2

11

21.40612841

Arizona

6

2

22.52961192

Ohio State

5

20

31.33341532

North Carolina

3

27

32.21368536

Kansas

9

12

33.03370355

Florida

13

3

34.15718706

UCLA

10

15

36.34279919

VCU

14

17

43.81113457

San Diego State

16

18

47.54530227

Maryland

8

40

47.60555411

Syracuse

21

16

54.28510028

Stanford

15

36

56.34910567

Georgetown

7

65

60.44699053

Michigan

29

10

63.4537239

Michigan St

31

9

66.03428216

Tennessee

32

7

66.45935422

Oklahoma St

26

26

67.94655489

North Carolina State

17

55

70.46575826

BYU

22

53

77.20555628

Xavier

20

59

77.5090216

Villanova

37

14

78.39039468

Providence

24

51

79.20930982

Indiana

19

67

80.54477784

UNLV

11

91

81.75863911

SMU

34

30

82.88322567

Seton hall

12

98

87.37495362

Utah

33

42

88.22620078

Oklahoma

38

33

90.92836578

Oregon

40

29

91.77850989

LSU

36

58

102.1911207

Arkansas

39

52

103.4663369

Purdue

23

97

104.1636453

Miami

35

69

106.9572911

Alabama

27

92

107.5943477

Vanderbilt

28

112

120.7091235

Northwestern

30

131

134.8257761

USC

25

163

145.3901195

Virginia Tech

18

192

151.0666859

Ok, so Maryland is now predicting to be the 12th best team in the country next year based on this (basic) model. Not bad. You'll notice that there may be teams that weren't in ESPN's top 40 Recruiting Class rankings, but had really good years last year, so maybe they would beat out Maryland if I had added them into the calculations for next year (like Wisconsin, Wichita St., Virginia). But there recruiting class ranking (being at best 41st) was so much higher than Maryland's that their finish this year doesn't make up for it according to my model.

Ok, so according to my model, Maryland's incoming recruiting class in addition to their performance last year has them expected to be the 12th best team in the country next year.

Again, let this not be a prediction or even a completely accurate expectation. Instead let's use this as a jumping off point. Let's talk and converse qualitatively about all the things that weren't included in the model that should affect these expectations and we can adjust these expectations based off of them. I suggest # of returning players, number of juniors and seniors, recent tournament experience, among some others. I haven't looked into the rest of the teams, but to me, Maryland's losses shouldn't hurt them too much and shouldn't push them down in the rankings too much relative to some other teams' losses.

I think some people might suggest the previous year's recruiting class should be factored in (that is, we should include the incoming recruiting class ranking AND the recruiting class ranking from the year before), but to me that already factors itself in with last year's finish. Also, some people might suggest the coach of the team. You should expect a better coach to finish better. Again, I think using last year's End of Season Ranking should factor in coach's ability and other things that carry over from one year to the next (obviously unless there is a new coach).

Hope you enjoy! Go TERPS!

Anything deemed inappropriate will be deleted by an admin or moderator with the power to do so. The views of the above FanPost do not represent the beliefs of Testudo Times or Testudo Times' authors, nor are they the work of them.

X
Log In Sign Up

forgot?
Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

By becoming a registered user, you are also agreeing to our Terms and confirming that you have read our Privacy Policy.

Join Testudo Times

You must be a member of Testudo Times to participate.

We have our own Community Guidelines at Testudo Times. You should read them.

Join Testudo Times

You must be a member of Testudo Times to participate.

We have our own Community Guidelines at Testudo Times. You should read them.

Spinner.vc97ec6e

Authenticating

Great!

Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.

tracking_pixel_9347_tracker