If anyone really cares anything about what Hollinger has to say...

[Dumpy]

<div class='quotetop'>QUOTE (ly_yng)</div><div class='quotemain'></p>

<div class='quotetop'>QUOTE (Dumpy)</div><div class='quotemain'></p>
<div class="ucomment">

He refuses to set forth the formulae he is using, stating only that we can find them in a book of his&ndash;so, it&rsquo;s hard to take an objectively critical view of his methods.</p>
</div>

</div></p>

http://www.basketball-reference.com/about/per.html</p>

</p>
<pre>uPER = (1/MP)* [ 3P + (2/3)*AST + (2 - factor*(tmAST/tmFG))*FG + (FT*0.5*(1 + (1 - (tmAST/tmFG)) + (2/3)*(tmAST/tmFG))) - VOP*TO - VOP*DRBP*(FGA - FG) - VOP*0.44*(0.44 + (0.56*DRBP))*(FTA - FT) + VOP*(1 - DRBP)*(TRB - ORB) + VOP*DRBP*ORB + VOP*STL + VOP*DRBP*BLK - PF*((lgFT/lgPF) - 0.44*(lgFTA/lgPF)*VOP) ]</pre>
<pre>factor = (2/3) - (0.5*(lgAST / lgFG)) / (2*(lgFG / lgFT)) VOP = lgPTS / (lgFGA - lgORB + lgTO + 0.44*lgFTA) DRBP = (lgTRB - lgORB) / lgTRB</pre>
<pre>pace adjustment = lgPace / tmPace</pre>
<pre>aPER = (pace adjustment)*uPER</pre>
<pre>PER = aPER*(15/aPERlg)</pre>
<pre></pre>

I definitely agree with your post, though. I'm happy with Hollinger in comparison to most other basketball "experts" out there, but I definitely think he could do better work.</p>

</div></p>

</p>

yoiks. This just stinks of regression analysis. What is "P?" [the first term in brackets] Anyway, it looks as though defensive rebounds are worth very little, and turnovers are worth about 50% more than assists.</p>

 

[lukewarmplay]

<div class='quotetop'>QUOTE (Dumpy)</div><div class='quotemain'></p>

<div class='quotetop'>QUOTE (ly_yng)</div><div class='quotemain'></p>

<div class='quotetop'>QUOTE (Dumpy)</div><div class='quotemain'></p>
<div class="ucomment">

He refuses to set forth the formulae he is using, stating only that we can find them in a book of his&ndash;so, it&rsquo;s hard to take an objectively critical view of his methods.</p>
</div>

</div></p>

http://www.basketball-reference.com/about/per.html</p>

</p>
<pre>uPER = (1/MP)* [ 3P + (2/3)*AST + (2 - factor*(tmAST/tmFG))*FG + (FT*0.5*(1 + (1 - (tmAST/tmFG)) + (2/3)*(tmAST/tmFG))) - VOP*TO - VOP*DRBP*(FGA - FG) - VOP*0.44*(0.44 + (0.56*DRBP))*(FTA - FT) + VOP*(1 - DRBP)*(TRB - ORB) + VOP*DRBP*ORB + VOP*STL + VOP*DRBP*BLK - PF*((lgFT/lgPF) - 0.44*(lgFTA/lgPF)*VOP) ]</pre>
<pre>factor = (2/3) - (0.5*(lgAST / lgFG)) / (2*(lgFG / lgFT)) VOP = lgPTS / (lgFGA - lgORB + lgTO + 0.44*lgFTA) DRBP = (lgTRB - lgORB) / lgTRB</pre>
<pre>pace adjustment = lgPace / tmPace</pre>
<pre>aPER = (pace adjustment)*uPER</pre>
<pre>PER = aPER*(15/aPERlg)</pre>
<pre></pre>

I definitely agree with your post, though. I'm happy with Hollinger in comparison to most other basketball "experts" out there, but I definitely think he could do better work.</p>

</div></p>

</p>

</p>

yoiks. This just stinks of regression analysis. What is "P?" [the first term in brackets] Anyway, it looks as though defensive rebounds are worth very little, and turnovers are worth about 50% more than assists.</p>

</div></p>

wouldn't it be points?</p>

</p>

 

[Dumpy]

you know, I'll bet "3P" = three-pointers made.</p>

 

[rory]

Maybe I'm reading it wrong, but are wins and losses anywhere in that formula?</p>

 

[pegs]

<div class='quotetop'>QUOTE (Dumpy)</div><div class='quotemain'></p>

<div class='quotetop'>QUOTE (GMJigga)</div><div class='quotemain'></p>

"douche"</p>

But you raise a fair complaint, and one that I agree with. I think the only means by which players should be evaluated are the ordinal counting scales and their closest derivations (FG, and then FG%, etc). It my belief that more complicated measures such as PER, +/-, or what have you, try to measure the "intangibles" in an attempt to offer more complete and conclusive analysis. However they fail in doing so; the complaints mentioned in this thread are just ones. I say lets err on the side of caution, and leave these "intangibles" out of quantitative measure. By definition, let us not try to measure the unmeasurable.</p>

</div></p>

Actually, I LOVE the plus-minus stat . . . when used to compare three, four, and five-man units, because it can indirectly measure the effectiveness of how teammates interact with each other. If, for instance, a combination of Kidd, Carter, RJ, Krstic and Collins has a higher plus-minus than a combination of Kidd, Carter, RJ, Krstic and Cliff Robinson, we can probably conclude that Collins intereacts better with those four teammates than Cliffy and should be utilized in that setting. It DOES not necessarily mean that Collins is better than Cliff, just that in some way his skills are being better utilized with that combination. Then we can recommend how Collins should be used in the future.</p>

Boki is a solid player, but I'm convinced that a combination of Krstic + Boki in the front court would NOT work very well, and using plus-minus in this way could show evidence of this. Then, we could start to develop ideas on how player rotations should be set up.</p>

</div></p>

Agreed with plus/minus. I love seeing the +/- on NBA.com (the Lenovo +/-) because it tells you what combos worked the best. I wouldn't be surprised if some coaches already used this in the NBA, it seems like a very useful stat.</p>

</p>

 

[pegs]

<div class='quotetop'>QUOTE (rory)</div><div class='quotemain'></p>

Maybe I'm reading it wrong, but are wins and losses anywhere in that formula?</p>

</div></p>

</p>

I don't believe so. PER is more of an individual stat, and wins/losses are a team stat, so I would guess that's why it wouldn't be included.</p>

While PER is flawed as is, adding wins/losses would probably make it worse, as it would probably make an average player like Gooden look as good as Elton Brand.</p>

 

[rory]

<div class='quotetop'>QUOTE (peg182)</div><div class='quotemain'></p>

<div class='quotetop'>QUOTE (rory)</div><div class='quotemain'></p>

Maybe I'm reading it wrong, but are wins and losses anywhere in that formula?</p>

</div></p>

</p>

I don't believe so. PER is more of an individual stat, and wins/losses are a team stat, so I would guess that's why it wouldn't be included.</p>

While PER is flawed as is, adding wins/losses would probably make it worse, as it would probably make an average player like Gooden look as good as Elton Brand.</p>

</div></p>

Hmm...OK, maybe it'd be better if they incorporated +/- stats into the PER. There's no way Zach Randolph can be a top 10 PF.</p>

</p>