Interesting Stats

dang bonzi is raw.

 

Hi, welcome to the world of basketball statistics.

The thing you have to understand about PER is that what's important is not how high it is, but rather HOW the players obtained that PER and trying to see what would fit better for the team.

Hayes' PER of 13.9 and Haslem's PER of 13.7 may seem similar but they got it in completely different manners. Hayes' 13.9 is obtained by a high rebounding rate, a high TS% (true shooting percentage that factors 3 point shooting and FT shooting), and a low usage rate. Haslem obtains 13.7 by an average rebounding rate, a decent TS%, and a high usage rate (showing more offensive competence). Now, for a team that has low usage rate players and lacks scorers, Haslem is a better fit. For a team that has high usage rate players and scorers, Hayes is a better fit.

PER also is biased towards more offensive players because the only defensive stat it factors is blocks and steals.

Feel free to ask any more questions.

 

Personally, I still tend to stay away from from stats like "PER" for role players because I feel it involves too many factors that shouldn't be put together. Instead I prefer to look at multiple stats that each concentrate on a single area.

This is very true in basketball, because role players can be more "specialized" - rebounder/defender, 3 point shooter, penetrator - whereas in baseball, even specialists do the same general things(getting outs and getting on base/driving in runs).

While basketball is indeed more team-oriented than baseball, it doesn't mean we can't improve upon traditional stats. For example, TS% and eFG% improves upon FG%. defensive/offensive efficiency are better measures of team offense/defense than points per game and points allowed per game.

 

This supports my strong belief that Bonzi should be starting at the expense of Shane. I'm sure Adelman is aware of this

 

how r u a member after 8 posts?!

 

Like Bonzi, I got skillz.

 

I actually wanted to make a new thread on this but because I can't yet...

People get turned off by Hollinger's PER (Player Efficiency Rating) because some people use it as an end-all be-all statistic. From all my experience with it, it's definitely not a perfect stat. For one it is biased towards offense because it only factors blocks and steals as defensive stats.

Now, given all that, PER became so popular because, despite all its flaws, it probably IS the best summary statistic out there. It's also very useful because it has the highest correlation to winning percentage than any other single statistic. It's correlation to winning is .85 (1 being "perfect", of course). For comparison's sake, the NBA's EFF statstic correlation is .66.

Needless to say, when one is evaluating teams statistically, a good place to start would be the PER distribution among players. I have chosen three great teams that I think are built similar to our new Rockets - the '94-'95 Rockets, the '00-'01 Lakers, and '06-'07

Here's a little rubric (albeit crude) to tell you in layman's terms what PER means (from wikipedia).

Code:

*A Year For the Ages:    35.0
*Runaway MVP Candidate:  30.0
*Strong MVP Candidate:   27.5
*Weak MVP Candidate:     25.0
*Bona fide All-Star:     22.5
*Borderline All-Star:    20.0
*Solid 2nd option:       18.0
*3rd Banana:             16.5
*Pretty good player:     15.0
*In the rotation:        13.0
*Scrounging for minutes: 11.0
*Definitely renting:      9.0
*On next plane to Yakima: 5.0

The cut-off to who gets shown is playing 500 minutes. Organized from most minutes to least.

The '94-'95 Rockets:

Hakeem Olajuwon - 26.0
Vernon Maxwell - 11.9
Kenny Smith - 16.2
Sam Cassell - 15.9
Clyde Drexler - 22.1
Mario Elie - 13.7
Robert Horry - 13.5
Otis Thorpe - 14.7
Carl Hererra - 11.7
Pete Chilcutt - 11.0
Chucky Brown - 12.4

The '00-'01 Lakers:

Shaquille O'Neal - 30.2 (!)
Kobe Bryant - 24.5
Rick Fox - 13.8
Horace Grant - 14.3
Isaiah Rider - 11.8
Brian Shaw - 10.9
Robert Horry - 11.0
Ron Harper - 12.5
Mike Penberthy - 11.2
Derek Fisher - 14.0
Devean George - 6.2
Mark Madsen - 9.2

The '06-'07 Spurs

Tim Duncan - 26.1
Tony Parker - 21.4
Manu Ginobili - 24.1
Michael Finley - 13.7
Brent Barry - 16.6
Bruce Bowen - 10.1 (I give him 3 more points because of defense)
Francisco Elson - 11.2
Fabricio Oberto - 11.9
Beno Udrih - 10.3
Matt Bonner - 16.4
Robert Horry - 11.9
Jacque Vaughn - 10.9

Projected '07-'08 Rockets (Last year's PER)

Tracy McGrady - 23.2
Yao Ming - 26.5
Rafer Alston - 12.9
Luther Head - 13.7
Shane Battier - 15.0 (I give him 3 more points because of defense)
Chuck Hayes - 13.9
Dikembe Mutombo - 14.4
Bonzi Wells - 8.9
Kirk Snyder - 12.2
-------Acquisition Line-------
Mike James - 13.3 (It was a huge down year from his previous 19.8, it'll probably go up with the Rockets)
Steve Francis - 15.1
Luis Scola - 16.3 (Hollinger Projection)

Distributions:

'94-'95 Rockets

25+: 1
20-25: 1
17.5-20: 0
15-17.5: 2
13-15: 3
10-13: 4
less than 10: 0

'00-'01 Lakers

25+: 1
20-25: 1
17.5-20: 0
15-17.5: 0
13-15: 3
10-13: 5
less than 10: 2

'06-'07 Spurs

25+: 1
20-25: 2
17.5-20: 0
15-17.5: 2
13-15: 4
10-13: 6
less than 10: 0

Projected '07-'08 Rockets

25+: 1
20-25: 1
17.5-20: 0
15-17.5: 3
13-15: 6
10-13: 2
less than 10: 1

Comment on the projected Rockets:

1. We're not going to have all these players
2. Bonzi Wells really should not have that crappy year again
3. Take Scola's PER with a grain of salt for obvious reason

I'm not going to do much analysis here, I just wanted to provide ya'll with the facts and you guys can make your own comparisons. All I'm going to say is that it's looking good for us :clap:

I hope I don't get attacked because of this like I did with the Haslem thread. I put in a lot of work in this because it's very interesting...

 

tmac's PER of 30.3 in 02-03 didn't get him mentioned as the "best player" in the L, which belonged even to kobe then. so it's all perception.

 

I would guess that when there are a lot of changes to a team (coach, personnel) from one year to the next, that PER of players will be less predictable. Also, PER can be inflated for players with high usage rates, and because Yao and McGrady will probably dominate possession usage for us some of those players who are usually high usage, low efficiency could see a steep decline in PER.

 

You're absolutely right. I admit I'm being very optimistic using last year's PER but one can dream. BTW, T-Mac's '02-'03 WAS one for the ages, highest ever by a guard not named Michael Jordan, if you look at that Magic team, it's a miracle they even made the play-offs.

 

because he's a coach

 

Matt Boner 16.4 ?????!!!!!

That explains why the Spurs let go of Scola and Butler, just to have enough money to sign Bonner (and Oberto)... I haven't seen much of the guy, but this is a shockingly high PER for a player so unknown...

 

Using the PW/PL (individual player wins and losses) instead of PER from the last two seasons, and my personal projection for minutes distribution of our team, I can make a (very rough) guess of how many wins we could have.

First, one guess for a breakdown of minutes that I wrote in another thread:

Of course, I don't expect we'll only play 9 players all season, but this is a rough guide to what we can expect on any given night. I'll assume that Butler is the primary backup C, and not Mutombo.

For each player, I'll look at their past 2 seasons, weighting 06/07 twice as much as 05/06. Here are the results:

Code:

                MIN     PW/48   PL/48   proj_PW proj_PL
McGrady         35      0.16    0.10      9.1    5.8
Yao             34      0.22    0.04     12.0    2.1
Battier         33      0.14    0.02      7.7    1.0
Francis         32      0.10    0.10      5.0    5.3
Scola (Juwan)   28      0.09    0.10      4.2    4.3
Wells           25      0.05    0.18      1.9    7.1
James           24      0.08    0.12      2.9    4.7
Hayes           17      0.17    0.01      4.8    0.3
Butler          12      0.06    0.13      1.1    2.6
                ---------------------------------------
                       Predicted Record:  49     33

I used Juwan as a replacement for Scola. Hopefully, Scola outperforms what Juwan gave us the last two seasons. This method might also underestimate the effectiveness of Francis, Wells, James, and Butler because (1) Francis and James played for some bad defensive teams last couple seasons which brought down their PW-PL record, and (2) Wells and Butler rated very poorly last year in somewhat limited playing time. On the other hand, I'm also assuming that the above 9-man rotation plays all 82 games -- which may inflate the final record.

On balance, I would expect between 50 and 55 wins with the current roster, if the minutes tend to breakdown as I projected.

 

From what I understand, using PW/PL (which is not that great of a stat to begin with, no offense) is not really appropriate for this kind of analysis. Your reasons for possible variation is right on the money though. A better way for this kind of analysis is to find estimated points allowed and scored using individual players stats and weighing with minutes (like you did) to make a total for a whole team and applying the Pythagorean Win Percentage on that difference. We don't have the means to do that yet because of our new acquisitions From last year, however, if Yao and McGrady stayed healthy, using my system (not really mine) projected winning 59-62 games. If we truly did improve this off-season (I think we did), get ready for one HELL of a season (assuming Yao and Tracy are healthy).

 

I'm not sure if that's any different from PW and PL. My understanding is they are essentially derived from ORtg ("points produced" per possession), DRtg ("points allowed" per possession), and possession usage.

 

I'm pretty sure it's different. In calculating PGm (to get to PW), you have to use estimates like stops (which is a huge estimate). Also, your methodology doesn't employ the pythag. The more I think about it though, your methodology is pretty solid too since my method (not really mine, again) uses estimates as well. I think the biggest problem of that particular analysis you did about the Rockets is Bonzi Wells (which you mentioned). Giving his horrible performance 25 mpg and all... Could you maybe try it with his Sacramento numbers?

 

Even though I am working in the finance field, I don't know much about basketball statistics, so please advise.

How do we find out the shooting percentage of player A when player A is guarding by player B? (like what's Kobe's % when Battier is guarding him)

Do we have any website to find out how often player A pass out the ball when player B is guarding him? (like when Battier is guarding Kobe, how often Kobe pass out the ball to someone else?)

Also website for when two players pair together, what are their total rebound, points, fouls and etc?... (Like Yao and Hayes)

 

Player Pair stats are at 82games.com. Examples:
http://www.82games.com/0607/0607HOUP.HTM
http://www.82games.com/0506/0506HOUP.HTM
...

For whatever reason, those pages aren't directly linked to. You just have to know the URL (add a "P" after the team's abbreviation).

The other ones aren't tracked anywhere online, as far as I know.

 

No, unfortunately I don't think so.

You can get something similar to the last one from,
www.82games.com

If it's not there, it's probably not anywhere.