well, just made a pretty big change to provide more information.. the "ERI" listed above is now known as "PERI" (Potential ERI), and ERI now represents a players index before the progression simulation. this image should help to explain:

NAC Players sorted by the new "ERI" (Pre-progression/Current)



any differences in ERI from previous screen shots (notably Truex), are a product of more recent index tweaks.. it's a work in progress, so these ERI ratings are subject to change as I further refine the process.

ok, time for sleep... more updates in the days to come

i think i cleared this up with you on msn coffey, but for those who are confused by my explanation:

the above screenshot is a list of NAC players sorted by the new ERI... the new ERI represents CURRENT effectiveness, before calculating progression. this means that seniors will naturally rank higher (or my index would be worthless). PERI represents a players effectiveness after simulating progression to that players senior season. if this list were sorted by PERI (by clicking on the PERI label in the program), you would see freshman, sophomores, juniors, and seniors all crammed at the top. when i get home from work tonight i'll post a picture of the same list sorted by PERI to help clear things up

Quote: Originally posted by elmore686 on 2/04/2008
For example, here is a list of the average ERI's for each class:

Freshmen - 265.03
Sophomores - 277.19
Juniors - 269.65
Seniors - 259.19

This suggests that the current sophomore class is "better" than the others in terms of their effectiveness as seniors.

Thomas did not recruit two seasons ago. This would fully explain why the sophomores stand out so much.

This post could not be converted. To view the original post's thread, click here.

This post could not be converted. To view the original post's thread, click here.

kujay,

i couldn't agree more.. thanks for pointing that out! there are definately some issues with weighting in the formula, and this is an obvious flaw. i appreciate your feedback a ton! i can't imagine why Mueller's PERI would be higher than Coder's, unless it has to do with the normalization of positions...

basically, after inputing all of the DIII players, i looked at average PERI's for each individual position. i then weighted each position to get these averages as close as possible. after looking at some of the data, it appears as though maybe i've unfairly weighted the PF position (more walkons in that group maybe?), so i'll have to take a look at it when i get home.

thanks again for the feedback! soon as i distribute the software and database to those who want it, you can easily look through the data and provide feedback such as this to further refine the formulas

the defense is accounting for the difference in the other stats for sure with hooker... blk isn't going to affect a guards ERI in any ammount worth noting

also, those stats above on class average ERI's are referring to all of DIII Allen, not just the NAC... something to chew on

After looking at the hooker/mueller comparison, it does appear that defense may be getting a bit too much weight

also worth noting.. the ERI formulas for PG and SG vary.. currently, a PG gets slightly more benefit from BH and PA ratings than a SG, while a SG gets much more out of PER ratings. i think those aspects may be alright at the moment (with minor tweaks needed), however, the defense weighting could definately use a tweak

So I was comparing apples to oranges. I think of Lawrence Mueller as a PG, but he is listed as a SG. I agree with how you have them set up too. I didn't realize that Mueller was being held to different standards than Hooker.

Still think the other Mueller shouldn't be ahead of Coder though ... even though they are different positions too.

Would it make sense to figure out ERI, PERI, etc. for all five positions? That's what I did with my silly little exercise. We have a few SG in our conference that play PG. And it seems like most of us interchange the PF/C quite a bit. And a bunch of guys can play SF; in fact, neither MMA or Thomas have a SF on the roster.

first, i agree on the Mueller over Coder 100%... i think i'll be able to fix that with a slight drop in PF weight (as i unfairly weighted it to start).

to answer the position question, i think in some instances it would be best to set the position of the player equal to the position they actually play (ie, set Lawrence Pueller as a PG).

the only differences in how a SG and a PG are rated are the ones i mentioned - PG's get slightly more out of BH and passing, while SG's get quite a bit more out of PER ratings.

for SF's, it's definately best to just take a player (PF or SG?) and treat them as a SF, as the ratings are done much differently (and rightfully so).

for big men, the differences in ratings are minimal.. in fact, i may just do away with two seperate formulas and treat them as one.. this way, the C and PF weights could be identical, and comparing them much easier

i remember an Adam Jackson/Reed Mendenhall debate a while back, and i thought i'd end any comparison swiftly.. here's what the HD Helper forecasts for each player at this point in their senior seasons:


Jackson-




Mendenhall-





Not to mention the fact that I'm pretty sure my algorithms are slowing progression too quickly (ie, mendenhall will be low 90's in BH and PA?)

oh, and ignore the FT ratings.. doesn't display those for senior year yet.

and i've made some important tweaks to the progression algorithm... much more realistic looking now across the board