My 33 pt method talent evaluation idea

Moderator: Doctor MJ

User avatar
Dr Positivity
RealGM
Posts: 59,689
And1: 15,484
Joined: Apr 29, 2009
       

Re: My 33 pt method talent evaluation idea 

Post#21 » by Dr Positivity » Fri Nov 2, 2012 8:03 am

It's not about how good the players are, it's a theory about what makes them that good, and that seems important to me (the "why" factor for both players and teams). It's the philosophy (1/3 physical tools, 1/3 skill and 1/3 feel for the game) that matters and not the numbers. By the way, the NBA isn't that good at evaluating talent. Otherwise they would draft players more consistently... or eg. they wouldn't be missing on how talented eg. Kevin Love, Greg Monroe, Stephen Curry are when they were drafted, when they got called "low upside/non all-star" prospects, which means that their talent was off the radar of the way they evaluate it, since those guys definitely have star talent

In terms of practical uses of this theory, the best way to test theories is always predictions. I've made them both for the rookie prospects, this year's college freshman and pros in their 1st to 3rd years in the NBA like Irving, Hayward, etc. I've also tried them out for teams (I don't trust applying it to teams nearly as much as individual players but I thought it was fun). I'm also planning on testing it against Vegas point spreads using those team rankings

In regards to a case like Valanciunas I really only see two options

- I am right about Valanciunas' talent level being mediocre

- Valanciunas ends up being a more talented player physically, skill wise or in feel for the game than I have given him credit for

If Valanciunas is a "20" instead of the "13" I ranked him as, I am 100% sure I will be able to explain why using those 3 categories based on the fact that I can explain every player's talent in the league using them
User avatar
AussieBuck
RealGM
Posts: 41,650
And1: 19,674
Joined: May 10, 2006
Location: Bucks in 7?
 

Re: My 33 pt method talent evaluation idea 

Post#22 » by AussieBuck » Fri Nov 2, 2012 11:09 am

Tony Allen's IQ is reprehensible on the offensive end. He looks to me like JR Smith would if he couldn't hit threes. He should hid from the ball like he's Biedrins.
emunney wrote:
We need a man shaped like a chicken nugget with the shot selection of a 21st birthday party.


GHOSTofSIKMA wrote:
if you combined jabari parker, royal ivey, a shrimp and a ball sack youd have javon carter
Rondo2Hondo
Banned User
Posts: 2,094
And1: 40
Joined: Dec 02, 2010

Re: My 33 pt method talent evaluation idea 

Post#23 » by Rondo2Hondo » Fri Nov 2, 2012 5:58 pm

Statistical analysis just took a giant step backwards, slipped on some moss, and fell down a well cracking it's head open in the process.

This is neither statistical evidence, nor analysis (little insight is given). I would be upset if we ever had the 33 point method show up on ESPN, or BBall-reference, haha
User avatar
Hendrix
RealGM
Posts: 17,030
And1: 3,662
Joined: May 30, 2007
Location: London, Ontario

Re: My 33 pt method talent evaluation idea 

Post#24 » by Hendrix » Sat Nov 3, 2012 5:03 am

In regards to a case like Valanciunas I really only see two options

- I am right about Valanciunas' talent level being mediocre

- Valanciunas ends up being a more talented player physically, skill wise or in feel for the game than I have given him credit for

If Valanciunas is a "20" instead of the "13" I ranked him as, I am 100% sure I will be able to explain why using those 3 categories based on the fact that I can explain every player's talent in the league using them


So basically, if you are wrong you are going to just change the scores 'after-the-fact'. What good is that? Basically all you are doing is putting a numerical grade to your opinion,and then when your opinion of a player changes, then your 'grade' changes to reflect your new opinion. This '33 pt method' isn't helping you "explain" why he would be good in this case.

You have Austin Daye rated as a 1 from a physical standpoint, and Durant as an 8, when they have a ton of similarities physically. If all of a sudden Daye started playing good would he all of a sudden have a good 'physical' score?

Changin scores is no different from Joe Blow saying 'I think James Harden is a fringe all-star' one week, and then the next week saying 'James Harden is a top 2 SG in the league'. In that case Joe Blow changed his opinion. In your case it would just be changing a number, which reflects your opinion, which is the same thing. James Harden hasn't magically changed as a player. He's the same player. He's just in a different situation which didn't show up in Joe Blow's opinion, nor would it show up in your numerical opinion.

In terms of practical uses of this theory, the best way to test theories is always predictions. I've made them both for the rookie prospects, this year's college freshman and pros in their 1st to 3rd years in the NBA like Irving, Hayward, etc. I've also tried them out for teams (I don't trust applying it to teams nearly as much as individual players but I thought it was fun). I'm also planning on testing it against Vegas point spreads using those team rankings

Well you are right that stats should be predictive. However, there is nothing anymore predictive in your '33 point' thing' than someones just general opinion. In the last paragraph you basically said that if your score wasn't representative of J Val, you would just change the score to represent your new opinion.

If Joe Blow evaluates 'x' prospect and makes a decision based on size, athletisism, ball handeling, shooting ability, ability to create, age, defense, college stats, etc... He is just forming an opinion based on what he's seen. With your method all you are doing is doing the exact same thing. Just forming an opinion about certain areas of the players game. The only difference is you are putting a # on those opionions, and plugging it into a fairly lazy weighted-average that doesn't have any reason behind it.

Please do not place bet using this. You are just betting based on your opinion on a select few areas of teams. It isn't much different then a regular sports bettor placing a bet based on their opinion of the teams. There's like a bazillion other things that factor into winning ball games.


Dr Positivity wrote:It's not about how good the players are, it's a theory about what makes them that good, and that seems important to me (the "why" factor for both players and teams). It's the philosophy (1/3 physical tools, 1/3 skill and 1/3 feel for the game) that matters and not the numbers. By the way, the NBA isn't that good at evaluating talent. Otherwise they would draft players more consistently... or eg. they wouldn't be missing on how talented eg. Kevin Love, Greg Monroe, Stephen Curry are when they were drafted, when they got called "low upside/non all-star" prospects, which means that their talent was off the radar of the way they evaluate it, since those guys definitely have star talent


There's nothing special about doing a lazy weighted average of your opinion.

Those players were all drafted in the lotto. Kevin Love for example was pretty much drafted 3rd overall. I doubt your formula would shed him in any higher of a light, as he (and the other 2) would probably not fair so well in the 'physical' area.



I really do not mean to be a dick here. But there is zero value in this. I would strongly advise you to stop putting time into this, because there is no value, and it is going nowhere. It happens. I put a decent chunk of time into this, viewtopic.php?f=344&t=1148517 , and ended up at the conclusion that this does exist, but there is insufficient current data to explain the 'curves' in real time, and it was basically a dead end without correct curves. I wasted some time, but **** happens.
oak2455 wrote:Do understand English???
User avatar
Dr Positivity
RealGM
Posts: 59,689
And1: 15,484
Joined: Apr 29, 2009
       

Re: My 33 pt method talent evaluation idea 

Post#25 » by Dr Positivity » Sat Nov 3, 2012 6:35 pm

You are misunderstanding what I am doing. This is not a stat or even a way to measure production. As it says in the title this is a way for me to grade talent, using a different perspective than the usual way talent is evaluated, putting a numerical basis to the idea that physical tools are way over-weighted in the evaluation of talent and feel for the game and skill are under-weighted. If my score for Valanciunas changes 3 years from now it will be because he has proven he's more talented than what I currently score him has. I do not see a problem with this in regards to the system, my score for him right now is an educated guess on how talented he is, that does not mean I'll get everything right, especially in the skill level category (while I am much more nature over nurture than everyone else in regards to skill, development does happen, if Valanciunas became a guy who hits 3s or like the best midrange shooter in the league he would plausibly score a 9 or 10 instead of the 7 I presently give him, for no other reason than "I guessed wrong on his skill potential"). If you are looking for an objective measure of players, this isn't it. This is grouping together my subjective evaluation of players' talent into 3 categories. If you think a system creating a "guide" like that is useless, that's fine, but I find it interesting to do. This is much more philosophy/theory than science, I am much more into the former in general but I know others prefer the latter

You also continue to talk like I start with a player's score (and thus my "total" opinion of a player") and divy it up between the 3 categories to match this number, when that's not what I do. I start with my opinions/grades of the players in the 3 categories and then add them up at the end. For example players that score lower than my previous opinion of them was because of this order include Dwight Howard, Josh Smith, Larry Bird, Russell Westbrook while James Harden, Gordon Hayward, Kyrie Irving, Hedo Turkoglu are 4 examples of players that came out shockingly high. For example I have Gordon Hayward as a clear all-star talent and I have Irving as one of the top 10-15 greatest talents of all time, and I had Harden as having like a 28 (superstar/Durant level) vs 21 (all-star) sized gap over Westbrook in talent level even before the Houston trade. Hedo scored like a 22 in my score which was shocking, the biggest thing I took from that is that Turk is one of the league's biggest enigmas and aside from his MIP year never came close to matching his talent in production. If the system was influenced by confirmation bias, there's no way Turk would come out with an outlier all-star score like that. Or for example most of last college season I had Thomas Robinson as a top 5 prospect like everyone else, but when I used this method he came out as like the 21st or 22nd best prospect in the draft and a future average player. At the same time Nicholson moved from a player I considered a 20+ pick guy to my 5th or 6th best player in the draft. The reason Robinson and Nicholson swung so much is that Robinson is elite in physical tools, middling in skill and poor in feel for the game, while Nicholson is poor in physical tools but elite in both skill and feel for the game, thus Nicholson excels in 2 categories vs Robinson's 1 and gets a much higher score. In these examples this was not me starting with an opinion and then making the system fit it, the system massively changed my opinion on these players. Likewise in my team rankings before I did them I assumed the Orlando Magic would be terrible like everyone else, but I am now predicting them to make the playoffs. The reason being I have them as #30 in physical tools, #1 in skill and above average in feel for the game, when added together they punch out as a middle of the pack playoff contender and not a bottomfeeder. Once again my previous opinion didn't effect this projection at all. The things I look for to evaluate a player's physical tools, skill and feel for the game don't depend on my 'final opinion' of the player. If you want to believe otherwise that's fine, but I personally know I am being unbiased when ranking players in these categories. I believe the system creating results that radically different from my opinions before using this system, is evidence confirmation bias is not influencing them

Also in regards to Austin Daye vs Durant, their physical impact on the game is not similar. The biggest thing I look for in physical score for perimeter players is how much of their offense relies attacking the basket and paint area. Durant isn't Lebron but he's still excellent at going to the rim and FT line compared to most SFs due to his athleticism/speed, and he's one of the best finishers at the rim in the league. Daye OTOH has no "attacking the basket" offense to his game at all, he's strictly 20 feet and out. Secondly Durant has a huge physical impact advantage on Daye defensively and on the glass, Durant is one of the best SFs in defensive physical and Daye one of the worst. Keep in mind in regards to the physical tools and skill scores, ballhandling has a bigger impact on the physical tools score than skill, and height has a bigger impact on the skill score than physical tools. This is because ballhandling helps a player attack the basket and height helps a player's shooting become more accurate. Because I am measuring the results (attacking the basket vs jumpshots, accurate shooting vs contested misses), not how they got there. Therefore aside from his explosiveness, Durant's ballhandling also helps him in the physical score vs Daye. The important thing is Durant imposes himself physically on the game way more than Daye. Daye is definitely a 1, Durant probably a 7 or 8, *maybe* someone could argue a 6. In regards to Durant, to add another point to your "subjective scores" argument, there's no way someone could argue him less outside of say a 24-29 range. Because Durant has to be scored as at least a 9 in skill and feel for the game vs SFs or that person doesn't know what they're talking about/are using a different way to grade players. And his basement physically has to be something like 6. While I have a hard time seeing anything higher than 9 physical 10 skill 10 feel for the game or something as justified for him. I don't know if it really matters if Joe Blow has Durant as a 24 and I have him as a 27, just like when a player has a 27 PER vs 24 PER people don't care about the difference, both are superstar scores. Likewise there's almost no subjective way to grade Westbrook over Harden in these categories. Westbrook has GOAT physical tools for a PG but average skill and feel for the game. Harden is elite in skill and feel for the game for a 2 and still top notch physically for a SG. That's going to put Harden over Westbrook on the total score every time even with subjectivity applied.
Gregoire
Analyst
Posts: 3,275
And1: 534
Joined: Jul 29, 2012

Re: My 33 pt method talent evaluation idea 

Post#26 » by Gregoire » Wed Jan 23, 2013 10:49 am

Great work, but why it did stopped? Its interesting to see how you rate such players like Lebron, Paul, Howard and all-time greats like MJ,Magic,Shaq,Hakeem ect
Heej wrote:
These no calls on LeBron are crazy tho. A lot of stars got foul calls to protect them from the league. The refs literally Nerf LeBron to protect the league. That's gonna be one of the most enduring takes for me from his career. :lol:
User avatar
Texas Chuck
Senior Mod - NBA TnT Forum
Senior Mod - NBA TnT Forum
Posts: 85,506
And1: 88,333
Joined: May 19, 2012
Location: Purgatory
   

Re: My 33 pt method talent evaluation idea 

Post#27 » by Texas Chuck » Fri Feb 15, 2013 2:17 pm

Gregoire wrote:Great work, but why it did stopped?


hopefully it stopped because he listened carefully to the points Hendrix was making. This is nothing more than Dr P's opinion on players not any kind of meaningful evaluation tool.. Way to much subjectivity to be of any value without Dr P first establishing meaningful credentials as a proven talent evaluator and even then it does nothing to establish a method of evaluating new players as they enter the league but rather one evaluator's opinion.

I too appreciate all the work he put into this and I hope he got some personal fulfillment from the project but it simply isnt what he believes it to be.
ThunderBolt wrote:I’m going to let some of you in on a little secret I learned on realgm. If you don’t like a thread, not only do you not have to comment but you don’t even have to open it and read it. You’re welcome.
Rondo2Hondo
Banned User
Posts: 2,094
And1: 40
Joined: Dec 02, 2010

Re: My 33 pt method talent evaluation idea 

Post#28 » by Rondo2Hondo » Sun Feb 17, 2013 10:52 pm

Gregoire wrote:Great work, but why it did stopped? Its interesting to see how you rate such players like Lebron, Paul, Howard and all-time greats like MJ,Magic,Shaq,Hakeem ect


Because what he was doing was nonsense.
hoop365dotcom
Ballboy
Posts: 4
And1: 0
Joined: Mar 08, 2013

Re: My 33 pt method talent evaluation idea 

Post#29 » by hoop365dotcom » Sat Mar 16, 2013 4:53 am

I think your 33 point evaluation makes perfect sense!
You've covered grounds for different "feats" on players. Smaller players are more susceptible to skills that smaller players should. vice versa with larger players. And then you have players like LeBron who simply can do it all on the court. Great post, i think their is some form of derivative in relation to how the PER is calculated and I think your post on this makes perfect sense to how statistical analysts come up with their numbers.
hoop365dotcom
Ballboy
Posts: 4
And1: 0
Joined: Mar 08, 2013

Re: My 33 pt method talent evaluation idea 

Post#30 » by hoop365dotcom » Sat Mar 16, 2013 4:55 am

Of course though, this is not a perfect formula, but to me it makes sense.
Do others agree?
Colbinii
RealGM
Posts: 31,359
And1: 19,556
Joined: Feb 13, 2013

Re: My 33 pt method talent evaluation idea 

Post#31 » by Colbinii » Sun Mar 17, 2013 4:53 pm

hoop365dotcom wrote:Of course though, this is not a perfect formula, but to me it makes sense.
Do others agree?

If you want an evaluation like this for all players buy 2k13...
tsherkin wrote:Locked due to absence of adult conversation.

penbeast0 wrote:Guys, if you don't have anything to say, don't post.


Circa 2018
E-Balla wrote:LeBron is Jeff George.


Circa 2022
G35 wrote:Lebron is not that far off from WB in trade value.
mmistras
Banned User
Posts: 122
And1: 5
Joined: Jul 12, 2012

Re: My 33 pt method talent evaluation idea 

Post#32 » by mmistras » Tue Apr 9, 2013 3:10 pm

I don't really get this. This is in the statistical analysis section but it has nothing to do with statistics. you're just subjectively allocating points from the eye test without any real basis for the points you're allocating. this isn't really useful until you can come up with a formula to back up the points you're giving players or everyone will disagree with each players' scores anyways
tsherkin
Retired Mod
Retired Mod
Posts: 78,413
And1: 19,948
Joined: Oct 14, 2003
 

Re: My 33 pt method talent evaluation idea 

Post#33 » by tsherkin » Wed Apr 10, 2013 6:50 pm

I think the method is a little less about hard statistics and a little more about having a semi-codified method of evaluation, like a scout's guide, that can be represented in numerical fashion. It's got a built-in system of scale and while that scale is subjective, it's an individual's opinion rather than any declared statement of fact.

If there were, perhaps, stat thresholds associated with skills, then this might proceed in a more quantitative fashion.

For example, you can look at finishing ability across zones, you can look at assisted percentage on those shots, you can look at overall eFG%, 3P%, FT%, Draw Rate, and you can compare all of those things in terms of standard deviations from league average...

Or if you want to get specific, you can compare them to league average for that position, perhaps within a given minutes requirement (30+ mpg, say, for starters). That way you can tier the skill based on number of standard deviations from positional league average. It's a lot of work, but it'd lend a little more statistical weight to this method.

Now, that doesn't work for something like "feel for the game," but you might make inferences on that basis from performance in certain things like rebounding, off-ball play, timing on blocked shots, charges, etc, etc.

With that kind of data integration, the 33-point method might become something a little more useful than an arbitrary assignment of scaled value and a list of traits to examine when evaluating a player.
User avatar
EvanZ
RealGM
Posts: 12,649
And1: 3,182
Joined: Apr 06, 2011

Re: My 33 pt method talent evaluation idea 

Post#34 » by EvanZ » Wed Apr 10, 2013 6:59 pm

It might be useful to use subjective ratings as a prior for RAPM or something like that.
I was right about 3 point shooting. I expect to be right about Tacko Fall. Some coach will figure out how to use Tacko Fall. This movement towards undersized centers will sweep ng back. Back to the basket scorers will return to the NBA.
tsherkin
Retired Mod
Retired Mod
Posts: 78,413
And1: 19,948
Joined: Oct 14, 2003
 

Re: My 33 pt method talent evaluation idea 

Post#35 » by tsherkin » Wed Apr 10, 2013 7:06 pm

EvanZ wrote:It might be useful to use subjective ratings as a prior for RAPM or something like that.


Honestly, I don't put faith into any single-metric evaluation of a player, since it can never substitute for a proper, broad spectrum analysis which includes video.

Something like this, though, it begins the process of evaluation rather nicely, especially if you don't reduce it just the grade, but include the actual data that feeds into the final result. Hot zones, versatility, maybe something like a range index, there are lots of things which could be extended from the type of data this would gather if it was done like I mentioned above (or at least following along the general principle, of course).

Return to Statistical Analysis