Hey guys, I’ve been working on adding a quantifiable preference function to my recruiting model and wanted to workshop some assumptions (because I don't think I have a strong grasp of preference multipliers!) The goal here is to arrive at a net preference advantage (or disadvantage) versus average competition for a wide set of data when combing through hundreds of recruits.

Method:

• Assign a numerical value to each preference possibility “Very good” = 1, “Good” = .5, “-“ and “ Neutral” = 0, “Bad” = -.5, “Very Bad” = -1
• Assign a weight to each preference (from PT to Longevity), adding more weight to more valuable preferences (i.e. Distance more valuable than Longevity)
  • Not assigning any value to PT given that is within the coach’s control
  • Assigning less value to success and longevity because I expect competition to be strong in those categories as well (again I am trying to find the expected value for ‘average’ competition)
• Take a dump of preference data to calculate the frequency of results, from this we can generate a numerical ‘Expected Outcome’
• Now compare the expected outcome against true outcome, the result is “Net Preference Multiplier”

Further Thoughts:

• The expected outcome is a product of my own dummy data, not my peer set. So, the ‘expected outcome’ is really my own expected outcome if a recruit has a preference in a certain category. Given I am not recruiting against myself, this may not be an accurate representation of competition.
• Is “Very Bad” more diminutive than “Very Good” (i.e. should weights be more variable?)
• Likely the best way to quantify preferences would be in the multiplier effect on APs, this would involve a few more steps, but may be a helpful check
• Results for particular preferences vary; a more accurate calculation would differentiate outcomes depending on preference possibilities (i.e. fast tempo vs perimeter vs strong D vs paint offense for “Play Style” rather than one overarching multiplier for Play Style)

Link:

• The below link details the calcs above which are relatively straightforward
• Testing area in the bottom half of tab 1 allows you to plug in a player # and gut check the formulas (we all know a good coach's eye beats Excel)
• Can make a copy of this sheet on your own to play around

Google Sheet

This is cool. You're probably underrating Success, Distance and Long-Term Coach

this is cool! word of caution - 'Likely the best way to quantify preferences would be in the multiplier effect on APs'

this assumes that the preference impact on unlocking / required AP to unlock, is the same as the preference impact on actual effort, yes?

i thought there was a good chance that would be true, but for recruiting actions. like, that the relative weight of ap to hv for unlocking, could be the same as the relative weight of ap to hv for actual effort. if this was true, it would be easy to get exact values for everything but CV. so i did exactly that, in my first couple months back to 3.0 (2 years ago now?). but the values are not the same, trying to apply the unlocking values of recruiting actions, to battles, yielded impossible results.

if seble took measures relating to the above (this may have actually come up as an issue in the beta, i'm kinda thinking i warned him about this back in the first place), so that unlocking consequences could not be used to reverse engineer recruiting effort, then i am thinking he probably did that for preferences, too. but maybe not. its still a worthwhile exercise, but, caution about reading into the results!

Posted by cubcub113 on 5/6/2022 12:30:00 PM (view original):
This is cool. You're probably underrating Success, Distance and Long-Term Coach

Thanks Cub! May be helpful to simply rank order the preferences from most to least impactful, how does the below feel?

[stealing a bit from Top's post recently in another topic]

Tier A
1. Wants to play
2. Distance
3. Success

Tier B
4. Longevity
5. Strong Conference

Tier C
6. Play Style
7. O / D

Posted by gillispie on 5/6/2022 1:17:00 PM (view original):
this is cool! word of caution - 'Likely the best way to quantify preferences would be in the multiplier effect on APs'

this assumes that the preference impact on unlocking / required AP to unlock, is the same as the preference impact on actual effort, yes?

i thought there was a good chance that would be true, but for recruiting actions. like, that the relative weight of ap to hv for unlocking, could be the same as the relative weight of ap to hv for actual effort. if this was true, it would be easy to get exact values for everything but CV. so i did exactly that, in my first couple months back to 3.0 (2 years ago now?). but the values are not the same, trying to apply the unlocking values of recruiting actions, to battles, yielded impossible results.

if seble took measures relating to the above (this may have actually come up as an issue in the beta, i'm kinda thinking i warned him about this back in the first place), so that unlocking consequences could not be used to reverse engineer recruiting effort, then i am thinking he probably did that for preferences, too. but maybe not. its still a worthwhile exercise, but, caution about reading into the results!

Interesting, definitely going to test this myself too. For sake of clarity, are you saying the multiplier effect of preferences had a different multiplier effect with recruiting effort (HV) as APs? Did you find a compounded effect on HVs given they are higher cost and more consequential on all in effort? Regardless of the multiplier, the effect should be directionally consistent (but understood on your warning to look out for the same multiplier).

Somewhat related...do you think preferences operate on a spectrum or are categorically effective. Much like how not all A+ prestiges are created equally, is every VG distance preference created equally? Although some preferences operate categorically (play style, O/D, PT), most seem like they should operate on a spectrum (will make quantifying a bit more challenging).

About 5 seasons ago another coach and I both went after a recruit, we both went all in AP's from the start. We were both A- prestige, we were both in the same conference, and we both run the same offense and defense. The only preference that was a factor was my opponent was listed as very good in perimeter offense and I was listed as very bad. After the first 2 cycles of recruiting it was obvious to me he had a advantage. Once I maxed out on promises, home visits, campus etc my opponent was very high and I was high. That proved to me how big of a difference very good compared to very bad can be.