The rating of individual players is subjective, so they obviously differ from site to site. Then, those player ratings are stuffed into a point calculation system that is unique to each site. Starting with different player ratings and then feeding those numbers into different calculation systems obviously produces different results on the different sites...sometimes drastically different.
Personally, I trust the guys at 247 to give the best recruit analysis. I'm not sure how their point system works, but since I trust the evaluations, I trust the outcome of the team rankings...even if it doesn't work out the best for Auburn this year.
However, there is something to how each site does it, so I thought I'd average them all out.
Since everyone uses different ranking/point systems, I took the number 1 team from each site and gave them a 1.00 point total. The #2 team was then a percentage of the #1 team. In other words, if the #1 team had 1000 points and the #5 team had 876 points, then the #1 team gets 1.00 and the #5 team gets .876. At 247, Alabama was #1 with a point total of 781. FSU was #2 with a point total of 773. 773 points is .989% of 781. So Alabama gets a 1.00 and FSU gets a .989.
I did that for all four sites, giving the top 25 teams on each site a % of the point total of the #1 team. The exception was ESPN, who doesn't show points. So I just subtracted .025 from each team counting down from #1. In other words, #1 was 1.00, #2 was .975, #3 was .950, #4 was .925, #5 was .900, and so on.
And here is the way they stacked up....
1. .985 - Alabama
2. .969 - Texas
3. .913 - Ohio State
4. .905 - Florida
5. .872 - Michigan
6. .854 - Florida State
7. .824 - Stanford
8. .815 - Miami
9. .796 - Georgia
10 .789 - Oklahoma
11 .772 - USC
12 .739 - Clemson
12 .739 - LSU
14 .715 - Auburn
15 .710 - UCLA
16 .703 - South Carolina
16 .703 - Texas A&M
18 .696 - Oregon
19 .636 - Tennessee
20 .615 - Washington
NOTES:
It was interesting to see certain teams all over the place in the rankings on each site. Here are some that had a wide range:
LSU -
#7 high on Scout
#17 low on Rivals
Georgia -
#5 high on ESPN
#13 low on Scout
Clemson -
#9 high on ESPN
#22 low on 247 Sports
South Carolina -
#12 high in Scout
#19 low on Rivals
USC -
#7 high in Rivals
#20 low on Scout
FSU -
#2 high on 247 Sports and ESPN
#10 low on Scout
Notre Dame -
#10 high on ESPN
#22 low on Rivals
Auburn's high and low was #12 on Rivals and #17 on 247 Sports and ESPN. And before anyone says "of course ESPN has us low", remember that they had us #3 and #4 the last two years...the highest two year average of any of the sites.
The truth is, recruiting rankings matter. They are an indication of the overall talent and depth you acquire each year. But the subjective nature of evaluating the players that make up the classes that make up the rankings means that the rankings should, as the very least, be viewed in groups. Is #7 really that much better than #12 on any given site? Especially since those two teams might be ranked #10 and #14 on another site?
In my opinion, I would group the top 20 teams by percentages related to the top team. Maybe the .900 teams, the .800 teams, the .700 teams, etc, as teams ranked that closely over the range of four different sites are essentially a subjective toss-up.
So for this year:
Group 1 consists of Alabama, Texas, Ohio State, and .Florida Group 2 consists of Michigan, Florida St, Stanford, and Miami.
Group 3 consists of Georgia, Oklahoma, USC, Clemson, LSU, Auburn, UCLA, South Carolina, and Texas A&M.
Group 4 consists of Oregon, Tennessee, and Washington.
In this case, I would only say there are eight classes that are CLEARLY ahead of Auburn's class. And that Auburn's class is on par, or in the same general ball park, as that of Georgia, Clemson, A&M, LSU, etc. After all, two sites had Auburn at #12 and #13, while two sites also had Georgia at #11 and #13. So even though Georgia averages out on the four sites at #9 and AU averages at #14, who's to say which ranking for each team are actually right. If you look at the two examples I just gave for each team, you'd think they were dead even.
This statistical analysis doesn't account for extraneous variables. Most notably missing is scholarship room. Auburn can't sign a full class this year unless we pull the "Saban" method and kick people off the team.
ReplyDeleteAlso, this analysis has low face validity and, under statistical analysis would probably have low concurrent validity and low internal consistency validity. A percentage is a poor way to quantify recruiting talent.
All recruiting ranking, this one included, have an additional problem. They fail to account for team needs. For instance, if a team has to pick up a fullback, they are usually no more than 1 or 2 stars. A team that recruits a fullback is at a statistical disadvantage in this system. It also doesn't account for prospects from small schools or prospects that haven't played very long and have serious potential.
Finally, your statistical system borderlines on statistical fudging. You used other scales (scout.com, rivals.com, espn) to provide your data. In order to run your analysis, you had to depend on the assumption that the other scales rate players accurately.
Don't be a bama fan. Rankings don't matter. We went undefeated in 2004 with average rankings. Many other teams have done the same without bringing in the number 1 or 2 recruiting classes every year. Everyone just needs to chill out. We filled all virtually all of our needs and brought in top players where they were currently needed. END OF STORY... War Eagle.
I know about the 247 formula but do not know how to work it. It has a sigma in it and I do not know how to work sigmas. Go to the 247 team rankings page and click on the "I" to the right of where it says team rankings. It explains the formula.
ReplyDelete