College recruiting of high school athletes attracts major attention – particularly in football and especially in Texas. Various Internet sites have emerged that rate and rank both individual players and school recruiting classes. The most popular of these sites are Rivals.com and Scout.com.
How accurate are their ratings, though? Obviously, the ratings are given based on high school performance, but they have the aim of predicting performance at the next level. Theoretically, those receiving 5-star ratings are more likely to "succeed" in college than those who receive fewer stars.
Now that websites such as Rivals.com and Scout.com have been around a few years, their ratings can be analyzed for accuracy. While only one of many possible measures of success, the various 2007 NCAA Division I All-American teams can be analyzed to see just how many of college football "stars" were identified coming out of high school.
2007 All-American Status
In order for a college football player to be considered an "All-American," he must be named to an All-America Team. In 2007, these teams were named by multiple institutions, only some of which were officially recognized by the NCAA in determining "consensus All-Americans" (marked with asterisks):
- American Football Coaches Association*1
- Associated Press*2
- CBS Sports3
- College Football News4
- Football Writers Association of America*6
- Pro Football Weekly7
- The Sporting News*10
- Sports Illustrated11
- Walter Camp Foundation*12
The various organizations differ in their approach to identifying All-Americans. Some named only first team selections, whereas others named first, second, third and honorable mention teams. The approach to special teams players also varied and only one team included a fullback position.
Given any team would likely be happy to have an All-American (AA) of any caliber on their team (whether he be given honorable mention or first team by one or all of the organizations listed above), data on all 2007 All-Americans has been collected in an online spreadsheet:13
The data includes the following information for each player:
- Star ranking received out of high school from both Rivals.com and Scout.com
- High school recruiting class
- All-America team status at each potential nominating organization (e.g., 1st team, 2nd team, etc.)
- Notes on position or school changes
Players who were not included in the recruiting analysis of a particular site were listed as having a zero star rating.
2007 All-American Data Analysis
The data is grouped by position in Tables 2 and 3 based on the ranking website. The number of each star ranking is listed for each position and for the 2007 All-Americans overall. The percentage distribution of each star ranking is also presented – by position and overall. Figures 1 and 2 display the distribution of star rankings within positions graphically.
Table 2: Compiled counts and percentages of the Rivals.com star rankings of 2007 AAs when they were high school recruits
Table 3: Compiled counts and percentages of the Scout.com star rankings of 2007 AAs when they were high school recruits
Table 4 presents a subset of Tables 2 and 3, listing only the average star rankings for each website, by position and overall. Figures 3 and 4 show the same graphically.
Figure 3: Average star ranking of 2007 All-Americans when recruited, by position and ranking website (bar graph)
Figure 4: Average star ranking of 2007 All-Americans when recruited, by position and ranking website (line graph)
Table 5 presents another subset of Tables 2 and 3, this time listing the number of players from each star ranking who were included in the 2007 All-America team rosters. Figures 5 and 6 are the graphical equivalents.
Table 5: Number of 2007 AAs receiving each star level when recruits and percentage of total number of 2007 AAs
While the data is interesting in its own right, real meaning is introduced when comparing it to what would be expected. In order to do this, more must be known about how Scout.com and Rivals.com gave out their ratings. For example, if Rivals gave out ten 5-star ratings and five of them became All-Americans, it would be much more meaningful than if Scout gave out fifty 5-stars and had seven "succeed."
Table 6 displays the number of 5-star rankings given out in each class with at least one member represented in the 2007 All-America team rosters, and the average over these years. The final column represents the percentage the number of 5-star recruits in the 2007 AAs (from Table 5) is of the total number of 5-stars given out in an average class (over the studied period).
While Scout had more 5-star prospects among the 2007 AAs, it also consistently handed out significantly more 5-star awards. Assuming the pool of players from which the AAs were picked is represented by an average recruiting class, Rivals had a higher percentage of their 5-star prospects earn AA status in 2007.
It would be of value, of course, to do the same comparison for the other star levels, but the necessary data isn’t available on either the Rivals or Scout websites, and neither organization was immediately forthcoming with the information.15 While some information is available on the subject, it does not provide a complete enough picture of the entire high school prospect class and how they are rated.16
If it is assumed the recruiting class star-ratings were normally distributed,17 the distribution of the ratings of those who succeed (such as the AA teams) should be "biased high" – if the ratings are accurate, at least. This trend may be manifesting itself in Figures 5 and 6, as there seems to be a slight bias to the right, but it is also apparent there is an abnormally large number of 0-star recruits present on the 2007 All-America teams.
Although it is tempting to draw conclusions from the data and analysis above, this is, in fact, only a one year analysis of All-Americans, which may not be representative of most years. Additionally, many assumptions were made along the way which must be taken into account. For example, while there are players who drop out, graduate early, get injured, are dismissed, etc., it was assumed an average of star distributions within recruiting classes represented the group of collegiate football players from which the 2007 All-Americans were picked.
Additionally, there are all sorts of factors influencing a player being nominated for an All-America team. Coaching, team roster depth and talent, academics, discipline, etc. – all of these could take a player who would otherwise have been a consensus All-American and land him working in fast food rather than preparing for the NFL. Nevertheless, the prospect rankings exist with the sole purpose of predicting the likelihood a player succeed at the generalized college level. Accurate rankings should take these factors into account and still show a much greater percentage of 5-star recruits making the All-America team than 0-stars.
See Part 2 of this article for further details and analysis.
1 "2007 AFCA Coaches’ All-America Team." AFCA_. November 29, 2007. Accessed January 2008 from http://www.afca.com/ViewArticle.dbml?DB_OEM_ID=9300&ATCLID=1329519ID=9300&ATCLID=1329519.
2 "Tebow, McFadden, Smith on AP All-America first team." Associated Press, ESPN. December 12, 2007. Accessed January 2008 from http://sports.espn.go.com/ncf/news/story?id=3150439.
3 "2007 CBSSports.com All-America Team." CBSSports.com. Accessed January 2008 from http://sportsline.com/collegefootball/story/10516777.
5 "Tebow, McFadden lead 2007 All-America team." ESPN_. December 8, 2007. Accessed January 2008 from http://sports.espn.go.com/ncf/columns/story? columnist=maisel_ivan&id=3158430ivan&id=3158430.
6 "FWAA Names 2007 All-America Team." SportsWriters.net. December 8, 2007. Accessed January 2008 from http://www.sportswriters.net/fwaa/news/2007/allamerica071208.html.
7 "Nolan Nawrocki’s All-America team." ProFootballWeekly.com. December 13, 2007. Accessed January 2008 from http://www.profootballweekly.com/PFW/College+Football/ College+Football+Extras/2007/allamerica.htm.
8 "All-America Team." Rivals.com. November 29, 2007. Accessed January 2008 from http://collegefootball.rivals.com/content.asp?SID=1144&CID=745101.
10 "SN’s 2007 college football All-Americans." SportingNews. December 7, 2007. Accessed January 2008 from http://www.sportingnews.com/yourturn/viewtopic.php?t=315722.
11 "SI.com’s All-Americas." SI.com. December 11, 2007. Accessed January 2008 from http://sportsillustrated.cnn.com/2007/football/ ncaa/12/10/all.americas/index.html.
12 "Walter Camp Football Foundation Announces 2007 All-America Team." WalterCamp.org_. December 6, 2007. Accessed January 2008 from "http://waltercamp.org/index.php/news/walter_camp_football foundation_announces_2007_all_america_team/":http://waltercamp.org/index.php/news/walter_camp_football_foundation_announces_2007_all_america_team/.
13 "College Football 2007 Div I All-Americans." Google Spreadsheet. Accessed January 2008 from http://spreadsheets.google.com/pub?key=pC3mpnNSv5yIeDY8bjgFPOA.
14 Where players were named to All-America teams at more than one position, they are listed as both. The more prominent recognition is listed first, and only this position will be considered in the analysis.
15 Please contact the author if you have access to this information.
16 DuMond, Lynch, Platania. "An Economic Model of the College Football Recruiting Process." Elon University. Accessed February 2008 from http://facstaff.elon.edu/jplatania/footballrecruiting.pdf. Researchers working with Elon University put together a very detailed analysis of how recruits decide what school to attend, studying the classes of 2002, 2003 and 2004. Table 1 in the Descriptive Statistics section at the end of the article include distributions of star rankings, but they apply only to the recruits studied, not the entire class.
17 "Normal Distribution." Wolfram MathWorld. Accessed March 2008 from http://mathworld.wolfram.com/NormalDistribution.html.
Similarly tagged OmniNerd content: