Stars - what they really mean!

[Brickhouse]

We need to stop star gazing! The stars arelazy way to evaluate the talent in a recruiting class. The real key is therivals rating system (see below). By in largewe are not going to get 6.1 or 6.0 guys. The last 6.0 guy we picked up wasLazard, guys like that are rare and they jump out as elite from day one……they don’tredshirt and they need to be home grown for the most part. Think Alabama andOhio State roster

My assessment it that the magic number is 5.7, Look backat our last 3 or 4 classes, sort by rating and take a look at the players with5.7 and higher………they are multi- year starters and high level contributors….forIowa State they are almost can’t miss guys. The 5.6 guys are mostly in the twodeeps but they are hit and miss on level of contribution – if they were on moreelite programs (OU, Baylor, LSU and TCU) they would rarely crack the 2 deeps. 5.5 guys and below are 50/50, half pan out andhalf don’t……these guys take a tremendous amount of time and effort to “developâ€â€¦â€¦atsome positions on the field you can develop them (OL) at other positions itsnot possible (especially where speed is concerned)………….the last 5.4 and 5.5guys that really worked out were Klein and Knott – was that development or didthe system just miss on these guys??? My guess if the rating system missedbecause it was hard to determine because they didn’t play against elite talentin HS

Bottom line is this, to compete in a P5 conference ittakes 5.6 and 5.7 guys and above………..our typical class needs to be composed of(1-2) 5.9 guys, (5-6) 5.8 guys and (3-4) 5.7 guys…………..the balance can be 5.6and below but to compete in the middle of the Big 12 year in and year out takesupper level talent….Our class needs to start looking like TTech, KSU, WVa…………….wecan’t be looking at the stars we need to be watching the Ratings…………don’t justbe excited about a 3 star guy, confirm that he is a 5.6 or 5.7 guy! Then get excited.

The ranking system ranks prospects on a numerical scale from 6.1-4.9.

6.1 FranchisePlayer; considered one of the elite prospects in the country, generallyamong the nation's top 25 players overall; deemed to have excellent propotential; high-major prospect

6.0-5.8All-American Candidate; high-major prospect; considered one of the nation'stop 300 prospects; deemed to have pro potential and ability to make an impacton college team

5.7-5.5 All-RegionSelection; considered among the region's top prospects and among the top750 or so prospects in the country; high-to-mid-major prospect; deemed to havepro potential and ability to make an impact on college team

5.4-5.0 Division Iprospect; considered a mid-major prospect; deemed to have limited propotential but definite Division I prospect; may be more of a role player

4.9 Sleeper;no Rivals.com expert knew much, if anything, about this player; a prospect thatonly a college coach really knew about

 

[ajk4st8]

 

[Mr Janny]

We need to stop star gazing! The stars arelazy way to evaluate the talent in a recruiting class. The real key is therivals rating system (see below). By in largewe are not going to get 6.1 or 6.0 guys. The last 6.0 guy we picked up wasLazard, guys like that are rare and they jump out as elite from day one……they don’tredshirt and they need to be home grown for the most part. Think Alabama andOhio State roster

My assessment it that the magic number is 5.7, Look backat our last 3 or 4 classes, sort by rating and take a look at the players with5.7 and higher………they are multi- year starters and high level contributors….forIowa State they are almost can’t miss guys. The 5.6 guys are mostly in the twodeeps but they are hit and miss on level of contribution – if they were on moreelite programs (OU, Baylor, LSU and TCU) they would rarely crack the 2 deeps. 5.5 guys and below are 50/50, half pan out andhalf don’t……these guys take a tremendous amount of time and effort to “develop”……atsome positions on the field you can develop them (OL) at other positions itsnot possible (especially where speed is concerned)………….the last 5.4 and 5.5guys that really worked out were Klein and Knott – was that development or didthe system just miss on these guys??? My guess if the rating system missedbecause it was hard to determine because they didn’t play against elite talentin HS

Bottom line is this, to compete in a P5 conference ittakes 5.6 and 5.7 guys and above………..our typical class needs to be composed of(1-2) 5.9 guys, (5-6) 5.8 guys and (3-4) 5.7 guys…………..the balance can be 5.6and below but to compete in the middle of the Big 12 year in and year out takesupper level talent….Our class needs to start looking like TTech, KSU, WVa…………….wecan’t be looking at the stars we need to be watching the Ratings…………don’t justbe excited about a 3 star guy, confirm that he is a 5.6 or 5.7 guy! Then get excited.

The ranking system ranks prospects on a numerical scale from 6.1-4.9.

6.1 FranchisePlayer; considered one of the elite prospects in the country, generallyamong the nation's top 25 players overall; deemed to have excellent propotential; high-major prospect

6.0-5.8All-American Candidate; high-major prospect; considered one of the nation'stop 300 prospects; deemed to have pro potential and ability to make an impacton college team

5.7-5.5 All-RegionSelection; considered among the region's top prospects and among the top750 or so prospects in the country; high-to-mid-major prospect; deemed to havepro potential and ability to make an impact on college team

5.4-5.0 Division Iprospect; considered a mid-major prospect; deemed to have limited propotential but definite Division I prospect; may be more of a role player

4.9 Sleeper;no Rivals.com expert knew much, if anything, about this player; a prospect thatonly a college coach really knew about

If that's the case, then this year is pretty underwhelming. Rhoads had more 5.6 and higher guys most of years he was here.

 

[jburke]

We need to stop star gazing! The stars arelazy way to evaluate the talent in a recruiting class. The real key is therivals rating system (see below). By in largewe are not going to get 6.1 or 6.0 guys. The last 6.0 guy we picked up wasLazard, guys like that are rare and they jump out as elite from day one……they don’tredshirt and they need to be home grown for the most part. Think Alabama andOhio State roster

My assessment it that the magic number is 5.7, Look backat our last 3 or 4 classes, sort by rating and take a look at the players with5.7 and higher………they are multi- year starters and high level contributors….forIowa State they are almost can’t miss guys. The 5.6 guys are mostly in the twodeeps but they are hit and miss on level of contribution – if they were on moreelite programs (OU, Baylor, LSU and TCU) they would rarely crack the 2 deeps. 5.5 guys and below are 50/50, half pan out andhalf don’t……these guys take a tremendous amount of time and effort to “develop”……atsome positions on the field you can develop them (OL) at other positions itsnot possible (especially where speed is concerned)………….the last 5.4 and 5.5guys that really worked out were Klein and Knott – was that development or didthe system just miss on these guys??? My guess if the rating system missedbecause it was hard to determine because they didn’t play against elite talentin HS

Bottom line is this, to compete in a P5 conference ittakes 5.6 and 5.7 guys and above………..our typical class needs to be composed of(1-2) 5.9 guys, (5-6) 5.8 guys and (3-4) 5.7 guys…………..the balance can be 5.6and below but to compete in the middle of the Big 12 year in and year out takesupper level talent….Our class needs to start looking like TTech, KSU, WVa…………….wecan’t be looking at the stars we need to be watching the Ratings…………don’t justbe excited about a 3 star guy, confirm that he is a 5.6 or 5.7 guy! Then get excited.

The ranking system ranks prospects on a numerical scale from 6.1-4.9.

6.1 FranchisePlayer; considered one of the elite prospects in the country, generallyamong the nation's top 25 players overall; deemed to have excellent propotential; high-major prospect

6.0-5.8All-American Candidate; high-major prospect; considered one of the nation'stop 300 prospects; deemed to have pro potential and ability to make an impacton college team

5.7-5.5 All-RegionSelection; considered among the region's top prospects and among the top750 or so prospects in the country; high-to-mid-major prospect; deemed to havepro potential and ability to make an impact on college team

5.4-5.0 Division Iprospect; considered a mid-major prospect; deemed to have limited propotential but definite Division I prospect; may be more of a role player

4.9 Sleeper;no Rivals.com expert knew much, if anything, about this player; a prospect thatonly a college coach really knew about

Though I agree that there have probably been just as many 4-5 star as lower rating that have exceeded. I look more at the schools that have offered a kid rather than his star ranking.... That tells a lot when we are beating out Michigan, Nebraska, Iowa, and Florida St for some of these kids, then the MAC school that majority of the recruits that CPR went after. I just love the energy and change that this staff has brought, next step is how do they coach this talent? Somewhere we had issues with in some places that past few years.

 

[iowa_wildcat]

Brick, have you counted the 5.6 and 5.7 guys that K-State has recruited since Snyder came back? Did you have to take your shoes off? Did you even need both hands? Snyder may be a HOF coach but a recruiter he is not. He simply coaches up the kids that have a fire in their belly.

 

[Acylum]

The space bar is your friend.

 

[roundball]

It's spelled "Starzzz"

 

[besserheimerphat]

Arbitrary Ranking & Scoring Evaluation (ARSE) A: Completely meaningless!

ARSE B: Totally the right way to do it!

How about we stop worrying about showing everyone that our ARSE is better than their ARSE and just let the coaches recruit.

 

[CYCLNST8]

The space bar is your friend.

Not if you're a droid...

 

[Rhoadhoused]

My advice is to not pay attention to rivals and to use the 247 composite rankings.

That's the average of Rivals, ESPN, Scout, and 247. It gives you a great average and their site is very easy to use.

They also don't have arbitrary, ridiculous cutoffs like Rivals with it's 6.1, 6.0, 5.9 nonsense.

It's a scale of 1-100, 4 decimal places. It's extremely user friendly and you can use their class calculator to add silent commits in or see what a certain recruit would do to your ranking.


http://iowastate.247sports.com/Season/2016-Football/Commits

For example, if Joshua Bailey and David Montgomery are silent commits you see our class goes from 50th in the country to 49th.

And since the scores are so specific, you could speculate on what Jacob Park might be worth if he was ranked at his composite ranking out of high school which was .9435.

You just have to get a kid really close to that and add him to our class. Do that and our class that improved to 49th goes up to 45th place. It's a fun tool, Trey did all this on the board today but I don't believe you need premium access to use the class calculator.


For the record Paul Rhoads' first class was 73rd by the 247 composite, and right now we're sitting at 50th. So it's clear to see Campbell is at a whole other level right now.

 

[besserheimerphat]

My advice is to not pay attention to rivals and to use the 247 composite rankings.

That's the average of Rivals, ESPN, Scout, and 247. It gives you a great average and their site is very easy to use.

They also don't have arbitrary, ridiculous cutoffs like Rivals with it's 6.1, 6.0, 5.9 nonsense.

But their numbers are based on aggregating those "arbitrary, ridiculous" values. And they assume that everything is linear - a 4 star guy is 2X better than a 2 star guy. Is that true? Does a 4 star add twice as much value to your team as a 2 star? All of these scales can give you an idea of the level of athlete for a particular player, but it can't realistically tell you how "important" that player will be to your program.

It's a scale of 1-100, 4 decimal places. It's extremely user friendly and you can use their class calculator to add silent commits in or see what a certain recruit would do to your ranking.

Decimal places is an indication of precision, not accuracy. Precision =/= accuracy

 

[Rhoadhoused]

But their numbers are based on aggregating those "arbitrary, ridiculous" values. And they assume that everything is linear - a 4 star guy is 2X better than a 2 star guy. Is that true? Does a 4 star add twice as much value to your team as a 2 star? All of these scales can give you an idea of the level of athlete for a particular player, but it can't realistically tell you how "important" that player will be to your program.



Decimal places is an indication of precision, not accuracy. Precision =/= accuracy

I have no idea where you got the idea that a 4 star is worth exactly twice as much as a 2 star or whatever.



And I never said decimal places equals accuracy and I know the difference. I said it makes it much easier to compare recruits and to use the system. It's user friendly. And you don't end up with 100 recruits at the exact same ranking of 5.7 or whatever.

The increased accuracy for me would be the 4 times larger sample size, the removal of outliers by averaging, the ability to go in and theoretically add guys who are silent commits or remove guys that are not going to qualify, ect.

The arbitrary, ridiculous values are arbitrary and ridiculous not because of the accuracy of the rating but because it's so counter intuitive. We have what, 8 levels that are stair steps and you can't be rated in between those? It's dumb. Not to mention the stupidity of those values being from 5.3-6.1 (?!?!) for some reason.

 

[besserheimerphat]

From the 247 website:

The 247Sports Composite is a proprietary algorithm that compiles prospect Rankings andRatings listed in the public domain by the major media recruiting services. It converts average industry ranks and ratings into a linear composite index capping at 1.0000, [emphasis added] which indicates a consensus No. 1 prospect across all services. The 247Sports Composite is the industry's most comprehensive and unbiased prospect ranking and is also used to generate247Sports Team Recruiting Rankings and Recruiter Rankings.

So it's taking that range for 5.3 to 6.1 (or 0 to 5 stars, or whatever else) and transforming it to a 0 - 1.0000 range and then averaging all of those transformations. Any of the step functions that you don't like are still in the new transformed value. But that gets hidden from view because the numbers are all averaged together. And because of the number of decimal places, I'm guessing that 247 shows differences in recruits that aren't really there. How many decimal places do you go before you say "these guys are the same?"

I'd rather see the transformed scores from each of the sources, as that would give an idea of how varied the opinions are of a particular player. For example, if a player is scored 0.8000 is that because everybody thought he was a 0.8000 player, or did the recruiting services range from 0.5000 - 0.9999 that happened to average 0.8000?

Regarding sample size - measuring one thing four times is not the same as measuring four things once. Measuring one thing four times is useful for determining how repeatable your measurement system is. But for recruiting, I think the goal is to determine where your players' expected value falls relative to other players' expected value. To do that you want to measure as many players as possible. Basically every potential D1 player has a profile on at least a few recruiting service websites, so practically speaking you are already getting a universal sample. You can't get any more samples because there aren't any.

 

[besserheimerphat]

Rhoadhoused, I don't mean to call you out specifically. I just see people everywhere that assume that because there is a number, it's legit and scientific and proven. Especially in the case of recruiting rankings, the numbers are typically nebulous and may not mean what people think they mean. I deal with this at work constantly when trying to communicate with management.

 

[kilroy]

I count 1 , 2, 3, 4, 5 ratings you have there and we all like to compare up to 5 stars hmmmm.

 

[cytown]

Go compare rivals 2011-2013 recruiting rankings then look at 14-16 on field results. See some patterns?

 

[Rhoadhoused]

From the 247 website:

[/I][/I][/I][/I]

So it's taking that range for 5.3 to 6.1 (or 0 to 5 stars, or whatever else) and transforming it to a 0 - 1.0000 range and then averaging all of those transformations. Any of the step functions that you don't like are still in the new transformed value. But that gets hidden from view because the numbers are all averaged together. And because of the number of decimal places, I'm guessing that 247 shows differences in recruits that aren't really there. How many decimal places do you go before you say "these guys are the same?"


You're making a lot of assumptions about the linear rating that they literally just mention in one sentence. And I still don't see how you can assume a 2 star is half the value of a 4 star. The 247 system really just goes from .7000 or so to .9999. It's not like a 2 star is a .25 and a 4 star is a .75.

And I think people can make their own judgements of how different recruits really are based on how far off they are in the 247 rating. Just use some common sense. This isn't a PhD defense. It's more intuitive than having thousands of recruits and only 8 different ranking levels, but they're still all ranked in a certain order without explanation. Rivals must have some system or detail that's more specific as well.




I'd rather see the transformed scores from each of the sources, as that would give an idea of how varied the opinions are of a particular player. For example, if a player is scored 0.8000 is that because everybody thought he was a 0.8000 player, or did the recruiting services range from 0.5000 - 0.9999 that happened to average 0.8000?

I just don't see why that would be necessary, just look it up yourself if it's that important. An average of the 4 services is fine. It's not as complicated as you make it out to be.


Regarding sample size - measuring one thing four times is not the same as measuring four things once. Measuring one thing four times is useful for determining how repeatable your measurement system is. But for recruiting, I think the goal is to determine where your players' expected value falls relative to other players' expected value. To do that you want to measure as many players as possible. Basically every potential D1 player has a profile on at least a few recruiting service websites, so practically speaking you are already getting a universal sample. You can't get any more samples because there aren't any.

Of course you aren't really measuring the same thing 4 times and getting the same answer. There's obviously a wide range of values that any recruiter can come up with watching the same tape. That value can be very high or very low in that range, there's now way to know. Having 4 separate evaluations helps to make sure that the rating reflects the industry consensus on that prospect. I think this is pretty obvious.

Your example of repeatability doesn't make much sense because we are using 4 different tools to measure the same thing 4 different times and we don't, in reality, know which of the 4 tools is the most accurate or what the real value of the rating is. And we can't ever know, really, without some advanced analysis beyond what anyone has ever done. Averaging the 4 different tools' results gives us a better idea of the true consensus rating of the prospect.

(I have no way to know which service is better at evaluations, so I'm assuming they're equal for simplicity sake.)

 

[Rhoadhoused]

Rhoadhoused, I don't mean to call you out specifically. I just see people everywhere that assume that because there is a number, it's legit and scientific and proven. Especially in the case of recruiting rankings, the numbers are typically nebulous and may not mean what people think they mean. I deal with this at work constantly when trying to communicate with management.

I understand more than most (while I'm no stats expert) I do have a STEM degree and have taken stats courses.

I'm just saying, this is football recruiting. We don't need to count out sig figs and all that stuff.

I think it's obviously better to use an average over the 4 recruiting services than to just cherry pick your favorite or whichever makes your team look good right now.

And maybe some people will see a guy listed at 87.45 and another guy at 87.46 and think that it means the 87.46 is way better, but I think we can apply some common sense and understand that they are similar. And as long as that extra .01 isn't weighted heavily in the 247 ranking system, it's not a big deal.

 

[BWRhasnoAC]

Rhoadhoused, I don't mean to call you out specifically. I just see people everywhere that assume that because there is a number, it's legit and scientific and proven. Especially in the case of recruiting rankings, the numbers are typically nebulous and may not mean what people think they mean. I deal with this at work constantly when trying to communicate with management.

Don't use facts when dealing with management. They get angry.

 

[tolfbfan]

The only rating I care about is the rating Campbell in his staff puts on a recruit and then their success in getting that guy. The rest is a money making venture for people who don't have real jobs!!:twitcy: