Rivals Formula

That makes more sense, because other sites are heavily influenced by the shear number of recruits.

Here's my interpretation:

The formula, H * ( n / ( n + m ) ) + L * ( m / ( n + m ) ), can be thought of in English as:

(Pts if the class is good)*(probablility the class is good)+(Pts if the class is bad)*(prob the class is bad)

(prob class is good)+(prob class is bad) = 1

They don't know the probabilities, but they do exist and they think it's better to acknowledge them than to ignore them. Maybe this is true, maybe not. It does give a boost to smaller high quality classes. However, the probabilities for most similarly ranked classes are likely pretty close, and I don't think the differences are really based on truth.

The guestimated probabilities used are variables dependent on the value of n. For the good classes at the top of the rankings, the prob the class is good most likely ranges from .8 to .9 (and for most teams in that group they're probably bunched closer together than this).

The prob the class is good goes up with the value of n. n increases for the following reasons:

o You've signed top 100 players: I can buy it. They've studied these the most and they're most confident in these.

o You've signed highly ranked players at their position: Conceptually, I'm fine with it, but I have a problem with how it's implemented. I do not understand how an ordinate ranking (1, 2, 3...) justifies the big boost in relative confidence that this gives. Being ranked #1 at your position will increase n more than any other factor (a star promotion will result in more net points, but that is not due to n increasing, but H and L). This seems ripe for manipulation and it has been in prior years.

For example, in an interview Rivals top rated center a couple of years ago, an AL commit who never played center nor is being recruited to play center, stated that a Rivals representive told him they moved him to center to boost AL's score. For example, if he was a member of the class in the example as an OT and you made no other change than switching him to center, the point total increased, not by a huge amount, but incomprehensible nevertheless.

o Sign 4 and 5 star athletes. Reasonable.

WRT the prob that the class is good, I just don't understand the point allocation to a players position ranking and I don't think the discontinuities artificially created at the top are rationally justified. I would opt to eliminate these because I don't think they buy you much.

WRT the points, I do not like the way they assign them as constants to the stars. This creates big discontinuities that have no meaning in real life. For example, a couple of years ago Tyler Love was #30 on the Rivals 100. He was a 5 Star (at the time 30 was the most 5 stars Rivals had named since they cut back in 2003. He was the 8th ranked OT. More than 1/4 of all 5 stars that year were OTs which are generally considered the hardest position to project future performance, but I digress.

What if this 30th ranked player was the highest ranked 4star instead of a 5 star? Bama would have lost something like 40 or so pts just because of this. On the 4 star/3 star end, it's probably around 20 points. How do you logically justify this? I don't think these kinds of discontinuities really exist. Personally I think they should assign points to smooth this out. 5 stars could average 250, but the pts should be assigned with a decreasing slope (#1 gets more than #30). Likewise with the other stars, so pts are smoothed out.

Bottom line is I think the formula is way too complicated and is ripe for manipulation.