Advanced Stat Primer - A Must Read for GSOMers
by FLAxwless on Mar 21, 2010 6:20 PM PDT
30 comments
So I'm editing this post because Jae has successfully answered my question and I didn't want to spam the boards.
I love talking hoops and I love debating about basketball, but most debates especially here on GSOM, usually come from two points of view: the arm chair scouts, those who rely on their observations/opinions and the "stat geeks", those who use stats as the end all be all. When these two sides get together to debate a player it generates a discussion along the divisive lines of: republicans vs democrats, religion vs atheists/agnostics, 2Pac vs Biggie. etc. Basically, it just gets plain ugly.
Like I said below I've been attempting to further educate myself regarding the strengths and weakness of all these advance stats that people are throwing around: PER, Wins Produced, Wins Shared, Roland ratings, and +/-. Many posts on GSOM cite these statistics in our discussions and there are those who swear by these statistics and those who simply just swear at these statistics. Luckily in my personal search for enlightenment I came across a series of blogs describing the what each of these statistics do, their strengths and weakness, and the proper applications of the stats. I found them so helpful and easy to understand that I am bringing it to GSOM in hopes that everyone can read these blogs, learn from them, and that we can all increase our knowledge and level of discussion.
So I encourage everyone stat geeks and arm char scouts to read these blogs by Andre "The Professor" Snellings that were published a year ago on rotowire.
Advanced Stat Primer
Can Math Help You Understand Basketball?
Hopefully, everyone can read these articles and we can increase the level of discussion regarding whether or not Anthony Tolliver is better than Reggie Williams.
----------------------------
I love using statistics as much as anyone and I've been trying to get a better understanding of some of the commonly used statistics. First up is John Hollinger's PER which many stat geeks do not like because of the arbitrary weights to certain categories and after reading into PER, I have come to value the stat less than WP and Wins Shared. While I agree with most of what D. Berri says regarding the flaws of PER in his blog, I am having trouble understanding this assertion that PER rewards inefficient shooting.
The same critique offered for NBA Efficiency also applies to Hollinger’s PERs, except the problem is even worse. Hollinger argues that each two rpoint field goal made is worth about 1.65 points. A three point field goal made is worth 2.65 points. A missed field goal, though, costs a team 0.72 points.
Given these values, with a bit of math we can show that a player will break even on his two point field goal attempts if he hits on 30.4% of these shots. On three pointers the break-even point is 21.4%. If a player exceeds these thresholds, and virtually every NBA played does so with respect to two-point shots, the more he shoots the higher his value in PERs. So a player can be an inefficient scorer and simply inflate his value by taking a large number of shots.
By solving the simple equations, I get very different percentages regarding the break even point:
X = FGM, Y = FGA
1.65x=0.72y for 2 pt FGs the break even point is roughly 43.7% vs Berri's stated 30.4%
2.65x=0.72y for 3 pt FGs the break even point is 27.7% vs Berri's stated 21.4%
So why is there a discrepancy? Not that this would make PER a superior stat to Wins Produced but I think it would give it more credibility when the break even point for 2 PT FG% is 43.7 vs 08-09 lg average of 48.5% and 27.7% for 3s vs 36.7% instead of the % that Berri is touting.
So can someone please explain to me the "bit of math" that Berri is talking about where he got those low % to break even?
This FanPost is a submission from a member of the mighty Golden State of Mind community. While we're all here to throw up that W, these words do not necessarily reflect the views of the GSoM Crew. Still, chances are the preceding post is Unstoppable Baby!
1 recs |
30 comments
|
Email
Print
Comments
Your values of x and y are not right.
X is FG made. Y is FG missed in Hollinger’s equation, not FGA. That’s the key.
I think
he just labeled y as FGA, because in the actual equation, it is set to FG missed.
Living for a magical 8th seed run
Labeling it as FGA is an error and the numbers he got reflect that error. If Y is set to be misses, as it is in Hollinger’s equation then the break-even point is indeed just slightly north of 30%. A 31% shooter who starts chucking will increase his PER.
DOH!
Something so simple. Thanks Jae.
Totally devalues PER for me now.
Check out Goallineblitz - Free Football MMORPG
Build players, Build teams, watch games...
http://goallineblitz.com/game/signup.pl?ref=4892220
Another question....
Where does Berri get the values for 2 and 3 FG made and the value for missed shots from Hollinger PER formula?
uPER = (1 / MP) *
[ 3P
+ (2/3) * AST
+ (2 – factor * (team_AST / team_FG)) * FG
+ (FT *0.5 * (1 + (1 – (team_AST / team_FG)) + (2/3) * (team_AST / team_FG)))
- VOP * TOV
- VOP * DRB% * (FGA – FG)
- VOP * 0.44 * (0.44 + (0.56 * DRB%)) * (FTA – FT)
+ VOP * (1 – DRB%) * (TRB – ORB)
+ VOP * DRB% * ORB
+ VOP * STL
+ VOP * DRB% * BLK
- PF * ((lg_FT / lg_PF) – 0.44 * (lg_FTA / lg_PF) * VOP) ]
Most of the terms in the formula above should be clear, but let me define the less obvious ones:
factor = (2 / 3) – (0.5 * (lg_AST / lg_FG)) / (2 * (lg_FG / lg_FT))
VOP = lg_PTS / (lg_FGA – lg_ORB + lg_TOV + 0.44 * lg_FTA)
DRB% = (lg_TRB – lg_ORB) / lg_TRB
I don’t see where they are getting that from.
Check out Goallineblitz - Free Football MMORPG
Build players, Build teams, watch games...
http://goallineblitz.com/game/signup.pl?ref=4892220
3P+ (2 – factor * (team_AST / team_FG)) * FG is where the values for 2 and 3 pt FGs made comes from.
- VOP * DRB% * (FGA – FG) is where the cost of a miss comes from.
Hollinger’s problem is that he does not take into account the opportunity cost of a possession in factoring in the first component. It is as if possessions are unlimited so long as you manage to make a basket on them, something that is not true.
Thanks.
Since those equations include team/individual variables such as AST and FG how is he able to come up with those constant values for possession? Is that a league average over a several year span?
Check out Goallineblitz - Free Football MMORPG
Build players, Build teams, watch games...
http://goallineblitz.com/game/signup.pl?ref=4892220
A 31% shooter who starts chucking will increase his PER.
Wow … I didn’t realize it was that bad. Sounds a bit like downgrading Bonds for taking 150 walks a season instead of swinging away like a man. Pretty shocking when you consider how long Hollinger has had to realize the error of his ways and adjust his metric accordingly. I have to wonder if he’s just an idiot, or an egomaniac, or a bit of both.
There will be no extra point!
by Sleepy Freud on Mar 21, 2010 9:27 PM PDT up reply actions
I highly doubt that
Can someone get on Hollinger’s next chat and ask them that question. I am across the ocean here otherwise I would do it myself. Thanks!
Hollinger has supported PER by saying that it coincides very closely with how people otherwise rate players. If that’s your measure, why would he want to change it? It doesn’t seem the least bit useful to me to have a measure that just seeks to put a number on what people have already determined to be true, but that is what PER seems to be best at.
Arguments like this one are the reason that some people doubt the power of statistics (not I, but others). Taking a preconceived notion and then creating a “metric” that reaffirms it is seriously lame.
Sittin in my scraper watchin Oakland goin wild, ta-dow!
Arguments like this one are the reason that some people doubt the power of statistics (not I, but others).
I don’t think that’s generally the case. I think most people doubt the power of statistics because they a) don’t understand them, b) have a math-phobia and/or c) find that the results do not mesh with their preconceived notions of how things are. I doubt very much that many stats doubters analyze the metrics and how they were created or tested closely enough to bother with seeing if they support is merely reaffirmation.
He does it to prove players like Kobe Bryant and Joe Johnson are great players, and likely top 10 players. Monta Ellis is an above average player this year for example…how?
Antoine Walker rated as an above average player throughout his career by JH, when we all know, he was a terrible, terrible player, who just took way too many shots.
Nellie needs to go! but while we suffer....John Wall or Evan Turner please!
Oh my god
Antoine Walker is so bad. Not a good rebounder, woefully inefficient scorer
by Reverend_Randy on Mar 24, 2010 10:17 PM PDT up reply actions
haha...
jae correcting someone’s stats.
by Cpt. Jack in the Box on Mar 21, 2010 8:24 PM PDT up reply actions
Is there a consensus?
Is there much consensus among the stat crunchers on this board as to a most superior evaluative system?
Can someone refer me to a site that lists WP and ORTG? DRTG column rankings for NBA players?.
by War Years Legacy on Mar 22, 2010 10:48 AM PDT reply actions
There is no clear choice. They each have their flaws.
These are the five I generally use:
http://www.basketball-reference.com has ORtg/DRtg, Win Shares and PER.
http://hoopdata.com/default.aspx has APER (adjusted PER that accounts for assisted field goals), and shot locations (a fantastic little thing). They also have motion charts, which are pretty cool. They have team and player data.
http://basketballprospectus.com/cardindex.php has a ton of things. WARP (wins above replacement), a different ORtg/DRtg (calculated to be the player with 4 teammates of average skill, aimed at being team-independent), as well as an assortment of defensive metrics that are too complicated to go into with a short blurb. Here are the explanations: http://www.basketballprospectus.com/article.php?articleid=518&mode=print
http://basketballvalue.com/index.php has adjusted plus minus.
http://82games.com has Simple Ratings, Opponent PER, FG% relative to shot clock position, 5-man unit effectiveness, opponent effectiveness (types of teams another team succeeds against, or struggles with), etc.
by Spider Jerusalem on Mar 22, 2010 4:59 PM PDT up reply actions
Oh, forgot about Berri.
http://dberri.wordpress.com has Wins Produced. Not listed anywhere specifically (other than end of season/mid season), but will do reports on players/teams periodically and update the WP for that franchise.
by Spider Jerusalem on Mar 22, 2010 5:01 PM PDT up reply actions
I love talking hoops and I love debating about basketball, but most debates especially here on GSOM, usually come from two points of view: the arm chair scouts, those who rely on their observations/opinions and the “stat geeks”, those who use stats as the end all be all.
You’re right. It’s just like politics: One one side you’ve got your bleeding heart liberals, those who want to spend hard earned tax payer money on every possible social development project rife with bureaucracy. On the other you’ve got your gun nut christian right wing conservatives, who will draw their NRA certified gun to kill anybody suggesting that abortion isn’t murder. There’s no middle ground at all. Nobody exists who sees both sides of the issues.
Other than the gross generalization, thanks for the links… though I suspect that the loudest and most ardent “arm chair scouts” will not click through to your links, nor would they understand them if they did.
Trust me, learning english isn’t a waste of time. It is actually sort of useful.
-randolphforpresident
by Dubs fan in Boston on Mar 22, 2010 1:16 PM PDT reply actions
As a “gross generalization”, I’ve found that those individuals who seem to give the best scouting reports based on their observations are also those individuals who are most open to utilizing statistics. I often feel like those who are rather adamantly against using statistics a) actually use a statistic and don’t realize they’re doing this, but that statistic is points-per-game and little more and b) often seem to be watching a completely different game than many of us as their observations just don’t match what I see with my own eyes.
Those who are in the middle are few and far between.
While we are on the topic of gross generalizations…. I feel that the “arm chart scouts” can easily understand the blogs that were posted as they are written for the audiuence and try to explain the what the statistic is all about.
Now getting them to click on the links is a different story.
Check out Goallineblitz - Free Football MMORPG
Build players, Build teams, watch games...
http://goallineblitz.com/game/signup.pl?ref=4892220
Re: I feel that the "arm chart scouts" can easily understand the blogs that were posted
Yah right. I tried to feed your fancy “statistics” into my Math Maker 2000 (shown below) and it just kept spitting out stuff like “Randolph needs beef”, “There’s no ‘eye’ in statistic” and “nuff said”.
I give pressure the reach-around.
by bloodsweatndonuts on Mar 22, 2010 4:02 PM PDT up reply actions 1 recs
HA! I have a Math Maker too! Mine got angry and chewed a hole in my wall when I rejected it’s “BBall IQ Hypothesis.”
Sittin in my scraper watchin Oakland goin wild, ta-dow!
No Way!!! I fed it a bunch of Skeptic’s posts yesterday and, for the next 6 hours, it kept trying to post me up in the low block.
Other statistical engines may return more “accurate” results but only Math Maker 2000 is equipped with state-of-the-art “killer clutch grit heart” technology that isn’t afraid to return wrong numbers with the game on the line.
I give pressure the reach-around.
by bloodsweatndonuts on Mar 23, 2010 10:08 AM PDT up reply actions
First order of business for Larry Ellison as owner...
Seek and destroy the GSoM Ladies’ Apparel Spammer?
There will be no extra point!
Wow
I really liked all of these articles, especially the one about Roland Rating. I had seen the stat, but didn’t quite understand it.
by Reverend_Randy on Mar 24, 2010 10:35 PM PDT reply actions
Glad to help.
Check out Goallineblitz - Free Football MMORPG
Build players, Build teams, watch games...
http://goallineblitz.com/game/signup.pl?ref=4892220
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2010 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
