11/26/2012 5:10 PM (edited)
Dahs, lets take that example ...

90 ATH, 5 LP

as a Low Post scorer I would have him at:

.3 * 90 + .45 * 5 =  27 + 2.25 = 27.25

Now a guy with 45 and 45 for both LP and ATH would be:

(.3 * 45) + (.45 * 45) =  13.5 + 20.25 = 33.75

So, the 45, 45 guy would be a higher LP scorer that the 90, 5 guy  ... but neither one is really "Good"
11/26/2012 5:10 PM
Posted by hughesjr on 11/26/2012 4:23:00 PM (view original):
Posted by coach_billyg on 11/26/2012 4:05:00 PM (view original):
Posted by hughesjr on 11/26/2012 3:33:00 PM (view original):
BTW, it is not just adding numbers together ... it is taking 100%, splitting that up into component parts, multiplying that percentage against the attribute, and adding those corrected values together.

By doing it this way, all your overall scores in any area will be in the range of 0-100 when comparing them to each other.
what you are describing is effectively, adding the ratings together. you are normalizing to a scale but thats not really the point.

to come up with the "right" formulas, dahs is right, adding definitely does not do the job. i wrote myself a program, similar to yatzr's but trimmed down, so i could rank recruits by custom formulas and recruit off it - although it was more about being able to play with the formulas, than actually recruiting (and i never really used it to recruit, but i played with formulas a **** ton). you minimally need cross products. for example, as we've been discussing, if you have good ath, lp is more valuable. if you have good spd, per is more valuable. so, to really capture the detail - you need an offense equation that would be something like-
3 point shooting = a*ATH + b*SPD + c*PER + d*BH + e*SPD*PER + f*BH*PER
Billy ... what is factor 'e' and factor 'f'

I am doing the a*ATH+b*SPD+c*PER+d*BH

Now if you effectively define (the way you are describing) a, b, c, d ... and if none of the other attributes matter, you would add those up and they would equal 100.

What would the e*SPD*PER do that you could not do with b*SPD and c*PER ?

Also think I may be doing the same thing by having 5 different "offensive types" ... in one type LP and ATH are very important and in the other type, PER and SPEED are more important.  Which ever number is higher (between those types) would be how that player is rated ... I'll post an example


i know what you mean, in your last statement, i was going to say something along those lines. we SORT of approximate it by saying for a slashing guard, ath/lp are more valuable, that kind of stuff. its a similar thing, but sort of in a different dimension.

so, heres the thing. for a 3 point shooting guard, for example, ath/lp are less important than other scoring guards, and spd/per are more valuable. but EVEN WITHIN THE SET of 3 point shooting guards - the value of spd and per vary together. those work off each other. so, to simplify this, lets assume spd = per in value for a 3 point shooter. let me give you an example. under that conclusion, a guy with 50 spd and 50 per, if you use a coefficient of 1 (i dont support the effort to put things on a 1 to 100 constraint and it gets more difficult when you add multiplicative factors, like e * SPD * PER, so i will not be normalizing to that range here) - anyway, using a coefficient of 1, that gives the player a rating of 100, from spd/per. well, 60 spd/per takes him to 120, and 70 spd/per takes him to 140. but that is not an accurate representation. the value of per is HIGHER when you have the spd to support it. instead of 50, 60, 70, 80, 90 spd/per being worth 100, 120, 140, 160, and 180 points, respectively, maybe its 100, 125, 155, 190, and 250. in reality, a 50 spd/per d2 guard isn't very good, and 60 spd/per isnt much better. a 70 spd/per guy can be pretty useful, an 80 spd/per guy can be a very good scorer, and a 90 spd/per scorer can be elite. that is a non linear relationship!

and that is really the key. so, to get around that, you need a different relationship. dahs suggests a quadratic on every variable, and multiplying them, but i am suggesting combining ratings and combinations of ratings, instead of your straight linear model.

so, to get back to the actual equation, say you use:
3 point shooting complex = a*ATH + b*SPD + c*PER + d*BH + e*SPD*PER + f*BH*PER
instead of 
3 point shooting simple = a*ATH + b*SPD + c*PER + d*BH

lets say in the simple model, you used:  .3*ath + .9*spd + 1.1*per + .7*bh:

well, in the complex model, you might use:  .3*ath + .7*spd + .7*per + .5*bh + (.4/50)*spd*per + (.4/50)bh*per, just as a simple example. you can also think of that spd*per term as .4*spd*per/50, or .008*per*spd

the reason I show the coefficient at .4/50 is to make it easier to understand. if i put that tiny coefficient, you might think, why is that even worth including? well, you are multiplying it by speed AND per, so thats going to be much larger values. if spd = 50, then that spd*per term is .4*per (same as .4*50*per/50). and really, that
(.4/50)*spd*per can be read, and maybe its easier for you to write it this way, as .2*per*(spd/50) + .2*spd*(per/50). its the same, but maybe easier to wrap your head around. if spd = per = bh = 50, then those terms basically break down from 
.2*per*(spd/50) + .2*spd*(per/50) + .2*per*(bh/50) + .2*bh*(per/50)
to
.2*per + .2*spd + .2*per + .2*bh

which, when you put it with the first part of the complex equation, is the same as the simple equation. so its saying that around the 50 level, the values of the ratings are the same as in the simple equation like yours. but at higher values, say per = spd = bh = 75, then its like having .3s instead of .2s in the above part, which makes spd, bh, passing all more useful as the ratings get higher. basically, the goal here is to encapsulate the reality that, the marginal value of per is both increasing in and of itself, but increasing with spd, too. the fact that the marginal value of per is increasing in and of itself, is the reason that dahs's use of exponents make some sense. you really probably get the best regression with something like this:
3 point shooting = a*ath + b*spd + c*per + d*bh + e*spd*per + f*bh*per + g*ath^2 + h*spd^2 + i*per^2 + j*bh^2    you might even make those squared exponents variables in the regression, but its the same idea...

anyway, getting back to it, using the simple equation and the complex equation:
simple - .3*ath + .9*spd + 1.1*per + .7*bh
complex - .3*ath + .7*spd + .7*per + .5*bh + (.4/50)*spd*per + (.4/50)bh*per

then if you have a player with 50 ath, and either 50/per/bh/spd, or 60, 70, 80, 90 - then here are the values he might have (note i was looking at a graph on google, values not exact, im rounding)
complex player ratings: 50 ath, X per/bh/spd
50: 150
60: 186 (+36)
70: 226 (+40)
80: 269 (+43)
90: 316 (+47)

you can control that to make different relationships, especially when you include exponents, you can really get whatever model you want. contrast that to the simple model:
simple player ratings: 50 ath, X per/bh/spd
50: 150
60: 177 (+27)
70: 204 (+27)
80: 231 (+27)
90: 258 (+27)

so, as you can see, in that model - you have significantly less flexibility. you have a linear relationship as values of ratings are increasing, but its not really accurate. i think its accurate to say, in d2, for a guy who sucks at spd/bh, going from 50 to 90 per is really not very useful at all, but for a guy with great spd/bh, going from 50 to 90 per is HUGE. you just can't capture that effect with the simple formulas we've been tossing around in this thread. still, that doesn't mean its not useful to talk about them - if you get too complex, it gets harder to follow, harder to wrap your head around, and fewer coaches can follow it. note that you can talk in just as much detail by saying, for a player like: 50 ath, 80 spd, 70 per, 70 bh - the values of the marginal ratings are, ath = .3, spd = .9 (1 point of speed is worth 3 points of ath), per = 1.1, bh = .7, or whatever you think it is. thats why in an earlier post, i was saying, well for a player with this set of ratings, here is how i value ratings, and for a player with that set of ratings, thats how i value ratings there.

the above lets you give a lot of detail in a way that people can understand, but it doesnt define what the relationships are at every stage. however, if you give a few different players and their ratings, i think people can extrapolate decently. on the other hand, its very difficult to come up with the equations that result in a curve that matches those relationships, and properly extrapolate in between - and further, if you saw that "perfect equation" - it would be really hard to think in your head, what that might actually mean for a particular player.
11/26/2012 5:11 PM
man it sucks im in a meeting, i type for 3 minutes and then it gets relevant, stop for 10, etc... i keep falling behind you guys, and my posts are getting outdated :)
11/26/2012 5:22 PM
Posted by hughesjr on 11/26/2012 4:59:00 PM (view original):
In my example ... (lets take Vanover) ... based on the 5 different "Offensive Types" I posted before (LP, Balanced, PE, SlasherPE, SlasherLP, SlasherATH) , his highest is PE and that is 72 ... so his overall offensive rating would be 72.

On defense, I now have 3 types, LP, PE, Balanced ... his best defense is PE, rated at 65.

So, basically there are 5 choices for an Offensive rating and 3 choices for a defensive rating.

Right now, I use these percentages for them:

Offensive Types ATH SPD LP PE BH IQ Total
LP 0.30 0.05 0.45 0.05 0.05 0.10 1.00
Balanced 0.20 0.10 0.20 0.20 0.20 0.10 1.00
PE 0.12 0.10 0.05 0.43 0.20 0.10 1.00
SlasherPE 0.30 0.08 0.05 0.17 0.30 0.10 1.00
SlasherLP 0.30 0.06 0.18 0.06 0.30 0.10 1.00
SlasherATH 0.38 0.08 0.03 0.03 0.38 0.10 1.00


Defensive Types ATH SPD DE BLK IQ Total
LP 0.17 0.02 0.50 0.16 0.15 1.00
PE 0.11 0.22 0.50 0.02 0.15 1.00
Balanced 0.12 0.13 0.50 0.10 0.15 1.00


I am not saying i am a genius or anything ... in fact  I am asking others what they think makes up a good PE Defense, LP Defense, maybe a Balanced Defense

In my example, a perimeter defense would be 0.11*ATH + 0.22*SPD + 0.5*DEF + 0.02*BLK + 0.15*DEF_IQ
and a LP defense would be 0.17*ATH + 0.02*SPD + 0.5*DEF + 0.16*BLK + 0.15*DEF_IQ

Again, in the case of Vanover, his perimeter defense is 65 but his LP defense is 56.

Billy, this is exactly what you did when you rated the PGs on the first page and said:   10% ATH, 30% SPD, 25% BH, 35% PASS ... if you took .1*ATH + .3*SPD + .25*BH + .35*PASS then that total would be YOUR rating for PGs.  The only difference is, I am creating 5 offense types and 3 defensive types.  I am also, like Trentonjoe, asking people what percentages of each attribute they think are important for each Position and also for each type ... am I missing some types, etc.

 





i know what you are saying hughes, i do something similar in yatzr's tool (which i believe you use). i only use linear equation because i was lazy and dont know where my complex forumlas are from my old program. but, ill fill out different roles, i might have 5 equations for a SG and if they are good in any of the 5 columns, i like them. so i definitely follow what you are doing, and i think its a good way to look at things. i did have some objections to your weights (ive mentioned some), but as a model, i like the style of saying a guy is the max of these 5 equations. as i explained in my last post, you can make more accurate equations if you don't use a linear equation, but its still definitely useful to do something like this with linear equations.

now, i know i am doing the same thing - i gave weights, that add to 100%. thats what you are doing too. however, thats the same as simply adding them together with a formula like .3*ath + .9*spd + 1.1*per * .7*bh, which does NOT exist on a 1-100 scale. when you look at your equations, you can derive the relationship between values, just like in mine, or in a different equation (like the ath/spd/per/bh one i just said) where you just add. and that relationship is always the same. so for example, in your post, for per defense, you suggest spd is twice as important as ath - or the ratio of the values of spd:ath is 2:1. thats an easy way to talk about this stuff - maybe for a pg, i think pass is 1.5x as important as bh, maybe you think its 1:1, etc... so your equation is just like mine (10% ATH, 30% SPD, 25% BH, 35% PASS ... if you took .1*ATH + .3*SPD + .25*BH + .35*PASS then that total would be YOUR rating for PGs) - where I say the value of spd to ath is 3:1. but its also just like a different one where you just add values of ratings, like  .3*ath + .9*spd + 1.1*per * .7*bh - where the value of ath:spd is 1:3, the value of per to bh is 1.1:7, or 1.57:1. 

so, i was just pointing out that when you told dahs you arent just adding ratings together - you really are, just like i have been all thread (until last post when i explained more complex, non linear formulas). really, it comes down to, are you using a linear formula or not - and you are, just like the OP did, and just like we all do normally. its a good way to talk about things, all dahs was saying (and hes right), is you cant get it all the way right there, like i explained in my last post.
11/26/2012 5:33 PM
I do agree that one might be able to possibly find a better curve than a linear one to approximate things ... but I also see value in keeping ratings at 0-100, etc.

The concept of yatzr's tool for both GD and HD, and the GD reports website for GridIron Dynasty, the Roles that WIS created for HD, etc. ... all do ratings that are based on a 0-100 scale and in a linear way.

I would certainly be willing to look at the other methods if someone would provide specific examples.
11/26/2012 6:52 PM (edited)
nobody is saying there isnt value in using a 0-100 scale (yatzr's hd tool doesnt work that way though, not for me at least). it makes it easier for some people to understand, thats good and all that. all we are saying is, that doesnt change anything material, its just perception. all you are doing is normalizing down to a 100 point scale, it doesnt change the ratios of attributes or give you a non linear methodology or anything materially different. its just like saying, i have values from 0 to 200, so you cut them all in half before you look at them, and now its 0-100. if that helps you, great, do it! a lot of people like to see it that way. but still, what you are doing is adding ratings together, and thats it - doesnt matter you are normalizing, not from the standpoint of accuracy of equations (this is in reference to your statement, "BTW, it is not just adding numbers together ... it is taking 100%, splitting that up into component parts, multiplying that percentage against the attribute, and adding those corrected values together." - this does not change anything about your equations or their accuracy, just the scale you are using them on, thats the point... it is just adding numbers together, like an equation that says 3 point offense = 3*ath + 6*spd + 10*per + 5*bh does (not that there is anything wrong with it)

i also gave you a very specific example in my earlier post, where i described a formula with spd*per in it. i gave examples of how it affects the recruits rating based on different values of the ratings. i dont think its an example of how i really weight those ratings, but its certainly an example of a non linear method. 
11/26/2012 7:33 PM
wrt yatzr's tool ... you are correct, it uses whatever math you include with the variables.  So it can do whatever is input.
11/26/2012 9:53 PM
The issue of what skill multiplier is best by position is an interesting question, but I think the idea of a looking at players in a vacuum is not the best way to go about creating a good team.  I look at 3 factors when deciphering how good a team is: 1) player skills 2) team cohesion 3) game planning.  I look at it as a 40%-40%-20% split in importance between the 3 factors.  Might matter differently between levels.  
I understand that people here aren't trying to ignore other factors, but I see a lot of talented teams that can't rebound or space the  floor and lose to inferior teams because they only focus on talent.
11/26/2012 11:27 PM
Posted by Trentonjoe on 11/26/2012 2:40:00 PM (view original):
"PGs - this one is the one i disagree with the most. for a triangle pg, spd and pass are the two most important ratings, and then bh. now, heres the tricky part, are you just talking the "PG ability on offense", which i often call "ability to run the offense", or just a PG as a whole? if you are just talking ability to run the offense, i would go with something like, 1.3x for pass to bh, and 1.1-1.2x for spd. hard to put that into 5%s but id roughly go with, 10% ATH, 30% SPD, 25% BH, 35% PASS"

All I really want out of my PG is to not turn the ball over. I think that is why I value BH (probably too much).

 Once that role is done, in my mind  he becomes either a primary scorer, a secondary scorer or a non scorer.  His passer rating importance is inversely proportionally  to his distrubution setting.    I think of PASS as a multiplier (similar to ATH), the higher the PASS rating the more smoothly the offense will run.  I also think that pass is more important to the people who are not shooting.   I can't imagine the SHOOTERS passer rating is factored into the "does the ball go in"  decision at all.

I'm pretty sure the PG's passing has more to do with their turnover rate than ball-handling does. Counter-intuitive I know, but that's the conclusion I drew from watching my Kentucky Wesleyan team last year. Tiny sample size I know, so maybe some seasoned players can weigh in. I asked another coach about it in a sitemail, which is still in my inbox, so here it is. Confirm or shoot down as you will. 

I was looking at stats, and it seems to me like passing is disproportionately important to turnovers, both for guys playing the point and for the off guards. Here are some relevant stats for my main guards (they've been rotating who starts until very recently, so quality of competition shouldn't be a confounding variable). 



Brown (PG)
53 ATH
75 SPD
70 BH
83 P
.9 turnovers per game (21.0 minutes per game)

Patten (PG)
39 ATH
62 SPD
94 BH
73 P
1.3 turnovers per game (19.3 minutes per game)

At this point, it looks like it might be a SPD/ATH issue overcoming Patten's vastly superior BH. But the SGs have a similar discrepancy without the matching SPD/ATH discrepancy. 

Lacy (PG playing exclusively SG)
57 ATH
75 SPD
57 BH
80 P
1.0 turnovers per game (20.7 minutes per game)

Ellinger (SG)
51 ATH
89 SPD
70 BH
57 P
1.8 turnovers per game (22.1 minutes per game)

I'm sure having completely awful BH would hurt, but it looks from this like passing ability is the most important factor in determining who can protect the ball. Or at least the most important factor in the triangle
11/27/2012 1:34 AM
billyg, rather than make a long quote string from a previous post, do you think among BH, PER, SPD, ATH, in your complex formula, that there is more than a two factor interaction?  ie .4/500 PER*SPD*ATH? or would you disregard that three and four factor interactions?
11/27/2012 9:41 AM
Posted by Trentonjoe on 11/26/2012 2:40:00 PM (view original):
"PGs - this one is the one i disagree with the most. for a triangle pg, spd and pass are the two most important ratings, and then bh. now, heres the tricky part, are you just talking the "PG ability on offense", which i often call "ability to run the offense", or just a PG as a whole? if you are just talking ability to run the offense, i would go with something like, 1.3x for pass to bh, and 1.1-1.2x for spd. hard to put that into 5%s but id roughly go with, 10% ATH, 30% SPD, 25% BH, 35% PASS"

All I really want out of my PG is to not turn the ball over. I think that is why I value BH (probably too much).

 Once that role is done, in my mind  he becomes either a primary scorer, a secondary scorer or a non scorer.  His passer rating importance is inversely proportionally  to his distrubution setting.    I think of PASS as a multiplier (similar to ATH), the higher the PASS rating the more smoothly the offense will run.  I also think that pass is more important to the people who are not shooting.   I can't imagine the SHOOTERS passer rating is factored into the "does the ball go in"  decision at all.

I went back and looked at finished seasons to see if there was an easily visible correlation between high PASS and low turnovers.    I didn't find one.

Turnovers to me are more related with FGA.  The more you shoot, the higher you turn the ball over.  I am sure there are other factors but this looks look like the biggest one.   

When comparing people with similar shot attempts, it seemed to me (at first glance, I haven't really run any numbers) that the lower BH guys had more turnovers.   I only looked at guards, BTW.

What I have noticed is that my low BH teams get killed by the press.

I really think PASS is used as a multiplier (or team PASS minus the shooter) in one of two equations in the possesion event tree.

1. Is there a turnover before a shot is taken?
2. Does the shot go in?
11/27/2012 12:30 PM
Posted by paynebrow on 11/27/2012 1:34:00 AM (view original):
billyg, rather than make a long quote string from a previous post, do you think among BH, PER, SPD, ATH, in your complex formula, that there is more than a two factor interaction?  ie .4/500 PER*SPD*ATH? or would you disregard that three and four factor interactions?
i would not disregard three and four factor interactions, it just gets increasingly hard to make sense of, and to freehand. you really need a graphing program to start to make clear sense of the situation.

in reality, i don't know what HD does for formulas, it could be anything, they have basically no restrictions on their end. it feels like there are combined terms in there - because as i was saying, it seems the jump from 50 spd/per to 60, 70, etc results in increasingly better and better players. but at some point, you start to tail off - a 3 point shooter with 90/90/90 spd/per/bh, compared to 95/95/95, compared to 100/100/100, i think the smallest jump in improvement is at that 100/100/100 level, which is really counter intuitive based on the increasing progression we see generally. i think what happens is this, HD programmer is basically looking at it going, well, an 80 spd/per/bh guy facing a 60ath/spd/def guy, hes got to do pretty well. what does he do, shoot 46% and 41% 3s? well, what does a 90/90/90 spd/per/bh guy do against the same competition? what about a 100/100/100 guy against a 40/40/40 defender? you cant really have guys putting up 70% and stuff, well, i guess you could, but it sure seems to me the HD developer didn't want that to happen. so, it seems to me, you really are hitting diminishing returns once you get up there, around 85 or 90. i remember OR saying, all players are basically the same once they have at least 85 in everything. i don't totally agree, but i think the effect he was trying to capture is the same as what I have seen myself.

assuming there is an increasing value in ratings up to, say, 85, and then it cuts off, its very possibly HD just uses different equations at different points of the curve. its ridiculously difficult for a user to correctly approximate that in a single formula, if thats the case. so we just have to do the best we can, without making it too complex, and in the end - there is no substitute for the eyeball test. for some people, that means they will just use the "simple" equations, which served me pretty well for pretty long, and some will try to do something more complex and maybe can make it a little more accurate. you definitely lose the ease of discussion though, thats why i am all for using simple equations for specific players or situations, with the understanding that different players in different situations have different equations.
11/27/2012 12:58 PM
tarvlon - you dont include FGA figures for those players, turnovers very much relate to FGA in this game. when we are trying to determine what factors are most important for something, we must always first figure out if we have all the factors - or we will reach the wrong conclusion.

trentonjoe - i definitely agree that pass is the more important rating for a pg. its not as much about his own turnovers, as the rest of the teams' turnovers. for your pg himself, bh is probably more important for turnovers - but if your pg is a non scorer, that bh figure just isnt that important. if your pg is a big time scorer, than that changes things. but for a traditional pg, who maybe is scoring you know, up to 8ppg or so, not really a dominant figure on your offense, his pass ratings is going to be significantly more important than bh, because his pass rating brings down the TOs of every other player on the floor, and increases their FG% (by getting them better looks)
11/27/2012 1:20 PM
Posted by gillispie on 11/27/2012 12:58:00 PM (view original):
tarvlon - you dont include FGA figures for those players, turnovers very much relate to FGA in this game. when we are trying to determine what factors are most important for something, we must always first figure out if we have all the factors - or we will reach the wrong conclusion.

trentonjoe - i definitely agree that pass is the more important rating for a pg. its not as much about his own turnovers, as the rest of the teams' turnovers. for your pg himself, bh is probably more important for turnovers - but if your pg is a non scorer, that bh figure just isnt that important. if your pg is a big time scorer, than that changes things. but for a traditional pg, who maybe is scoring you know, up to 8ppg or so, not really a dominant figure on your offense, his pass ratings is going to be significantly more important than bh, because his pass rating brings down the TOs of every other player on the floor, and increases their FG% (by getting them better looks)
Didn't even think of that. And it was last season, so I don't have the data anymore. 
11/27/2012 1:28 PM

"trentonjoe - i definitely agree that pass is the more important rating for a pg. its not as much about his own turnovers, as the rest of the teams' turnovers. for your pg himself, bh is probably more important for turnovers - but if your pg is a non scorer, that bh figure just isnt that important. if your pg is a big time scorer, than that changes things. but for a traditional pg, who maybe is scoring you know, up to 8ppg or so, not really a dominant figure on your offense, his pass ratings is going to be significantly more important than bh, because his pass rating brings down the TOs of every other player on the floor, and increases their FG% (by getting them better looks)"


I don't disagree with most of this, but how can you quatify it?   I am starting to think a true PG is the hardest position to recruit.   You need ATH, SPD, DEF, BH and PASS.
of 4

Terms of Use Customer Support Privacy Statement

Popular on WhatIfSports site: Baseball Simulation | College Basketball Game | College Football Game | Online Baseball Game | Hockey Simulation | NFL Picks | College Football Picks | Sports Games

© 1999-2014 WhatIfSports.com, Inc. All rights reserved. WhatIfSports is a trademark of WhatIfSports.com, Inc. SimLeague, SimMatchup and iSimNow are trademarks or registered trademarks of Electronic Arts, Inc. Used under license. The names of actual companies and products mentioned herein may be the trademarks of their respective owners.