Someone explain Rivals ranking system to me

[AndyDufresne]

I don't understand Rivals rankings. The very base measure of a recruit is the stars. Now one would assume that a 4-star is twice as likely as a 2-star of finding success on the field, given basic mathematics. Now Rivals throws out their numerical ratings. The highest rated 2 star is given a 5.4. The lowest rated 4 star is given a 5.8. Using these numbers, the 4 star is only 1.074 times more likely to succeed. What does Rivals see as the true measure of value? Why use both units of measure? Why start a 2-star at a 4.9 and end with a 5-star getting a 6.1? Seems kind of odd starting and stopping points to me.

 

Add to this the team rankings, which I don't understand at all. Last year, Oregon and Notre Dame both had 23 recruits. They each had 10 four stars, but Notre Dame had 12 three stars and 1 two star while Oregon had 1 five star, 9 three stars, and 3 two stars. They both had the exact same star average, and the total of the numerical ratings (5.5, 5.8, etc.) both were 131.2. And yet Oregon's team ranking is 233 points higher in the team rankings, or nearly 15% higher than Notre Dame's. With things being so equal, why the disparity?

[1995 Huskers]

read the description of a 5.4, 5.5, 5.6, etc. on rivals go next to the horizontal yellow bar and click About

[AndyDufresne]

read the description of a 5.4, 5.5, 5.6, etc. on rivals go next to the horizontal yellow bar and click About

 

I have read the descriptions and understand what the numbers represent. I am more interested in the reasoning for the assigned values and calculations, which I'm not sure anyone not affiliated with Rivals really knows. I'm hoping someone has some insight.

[zoogs]

Some interesting reading

[AndyDufresne]

Some interesting reading

 

Thanks for that. I also found this: Rivals calcs

 

Looking at the formula, it is interesting that so much of the team ranking depends on the "N" variable, which only awards bonus points for position rankings and being in the Rivals 100 (or Juco top 50). Being in positions 101-250 of the Rivals 250 is basically meaningless, as are the numerical ratings (4.9-6.1). Kenny Bell was the #67 athlete and a 5.7 three star last year, but if the formula stands and bonus points are only awarded to the top 25 athletes he was worth less points than several players that were rated as 5.6 three stars because they were within some arbitrary number for position rankings.

 

I was surprised that the formula was developed by one statistician from Cal. I would have guessed that they would have put more time and effort into developing the formula than to hire one guy to come up with a model for them still using position ranking cutoffs developed by Jeremy Crabtree nearly a decade ago.

[haymarkethomer]

The easiest way to explain the rankings is...

 

USC/Texas/Florida/Alabama etc Commit > Nebraska Commit

[Rosh]

Another way to look at it is this:

 

2 felony convictions = Miami Commit

1.1 GPA and a 9 ACT = Alabama Commit

2 Shoplifting convictions = Florida State Commit

Just Plain Dumb = USC/Lame Kiffin Commit

 

Excellent work ethic/stats to back it up/Intelligent = 3 star Nebraska Commit

[kchusker_chris]

Looking at the formula, it is interesting that so much of the team ranking depends on the "N" variable, which only awards bonus points for position rankings and being in the Rivals 100 (or Juco top 50). Being in positions 101-250 of the Rivals 250 is basically meaningless, as are the numerical ratings (4.9-6.1). Kenny Bell was the #67 athlete and a 5.7 three star last year, but if the formula stands and bonus points are only awarded to the top 25 athletes he was worth less points than several players that were rated as 5.6 three stars because they were within some arbitrary number for position rankings.

 

I was surprised that the formula was developed by one statistician from Cal. I would have guessed that they would have put more time and effort into developing the formula than to hire one guy to come up with a model for them still using position ranking cutoffs developed by Jeremy Crabtree nearly a decade ago.

I posted the calculation as a procedure last year sometime, you can probably search for it. I think you can google for excel spreadsheet's that are already set up for you as well.

 

When the classes are smaller, the average star ranking has a much greater affect on the class calculations than on classes 20+. As of now, adding a 2 star really hursts us, adding one later in the year has a smaller impact. A lot of the calculations assume larger class sizes, which is why they don't post team rankings until later in the year.

 

I personally like the concept behind the algorithm, however I don't think it's been updated for a while. The position numbers...

 

'DQB',25

'PQB',25

'RB',35

'FB',15

'WR',50

'TE',20

'OT',40

'OG',30

'OC',10

'DT',50

'WDE',20

'SDE',30

'ILB', 35

'OLB', 35

'CB',40

'S',30

'A',25

'K',5

 

...might relate to their scouted players 10 years ago, but not today - when they are averaging 10-15% more ranked players each year. They need to increase some of those numbers. I also don't like that a JUCO #99 player, gets the same points as a Rivals #99 player. The difference is the JUCO player probably doesn't make the position ranking, or contribute as many stars to the average.