FanPost

Comparison of PER and WS/48 and how it relates to the draft

With all the player profile reports I have been posting, I am sure that you notice that I have placed a strong emphasis on some very specific metrics. The most important metric's I refer to regularly are WS/48, Ortg, DRtg, and PER. When I do my player rankings I will use each of these as a part of the way I rank players.
My analysis places significant emphasis on WS, and Ortg and DRtg, less emphasis on PER, though I do consider PER. I recently did a study comparing WS and PER to better understand both metrics, and the results greatly influence how I evaluate players.
What I have done with my analysis is to compare and contrast PER and WS. How I did this was first to take all NBA players, and sort out the low minute players, and then rank the remaining players to rank the players best to worst in PEr and also best to worst in WS/48min, and I did this for 4 seasons. The 3 highest rated players using PER, in order, were LeBron James, Kevin Durant, and Chris Paul. The 3 highest rated players using WS were, in orde,r exactly the same as with PER, LeBron James, Kevin Durant, and Chris Paul. So in some instances WS and PER give the same results. In many case though the results were quite different.
For players where PER rates a player higher than WS, the top 5 players in terms of positive PER deviation were in order; 1.) Jrue Holiday (PER 92, WS 281), 2.) Kemba Walker (PER 44, WS 234), 3.) Eric Gordon (PER 133, WS 301), 4.) Chris Kamen (PER 107, and WS 274) and 5.) Josh Smith (PER 66, WS 232).
For players where WS rates a player higher than PER, the top 5 players in terms of positive WS deviation were in order; 1.) Quincy Pondexter (PER 278, WS 99), 2.) Shane Battier (PER 286, WS 111, 3.) Udonis Haslem (PER 314, WS 143), 4.) Matt Bonner (PER 251, WS 82), and Thabo Sefolosha (PER 219, WS 58) (note there were a couple of other players in this general group with lower minutes in is group WS top 5 that I am not showing, because I wanted you to see players who played similar minutes).
So how do you determine which tool is better, or are they essentially similar, just wrong in different ways? What I did was break the population (about 340 players each season) into three groups. Group #1 is the players where PER and WS/48 rank the players in a similar manner. I made this group 50% of the population. Group #2 are the players where the differential between PER and WS was the greatest in the favor of players rated highly by PER, and I included 25% of the population in this group. Group #3 are the players where the differential between PER and WS was the greatest in the favor of players rated highly by WS, and I included 25% of the population in this group.
For these 3 groups, I found the general nature of the groups. I applied the number of wins their actual NBA team had, so that I could find the relationship between actual NBA wins, and PER and WS. I calculated the average stat line for the players in each of the groups, including things such as ORtg, DRtg, TS%, eFG%, rebounding, assist, steal, block, and turnover rates, as well as the players usage rate.

Season Measurement Category Avg Team Wins Winning % PER WS/48 TS% eFG% TRB% AST% STL% BLK% TOV% USG% ORtg DRtg
2012/13 Winning % for Players where WS is HIGH and PER is LOW 49.4 60.3% 12.4 0.114 55.2% 52.3% 10.2% 9.7% 1.6% 1.8% 13.8% 14.6% 110 104
2012/13 Winning % for Players where WS and PER are consistent 41.4 50.5% 14.6 0.097 52.5% 49.1% 10.0% 14.1% 1.7% 1.7% 13.3% 19.5% 105 106
2012/13 Winning % for Players where WS is LOW and PER is HIGH 30.0 36.6% 15.1 0.067 51.4% 47.6% 9.7% 18.0% 1.6% 1.5% 13.7% 22.6% 102 108
2012/13 Additional wins for WS over PER players 19.4
2011/12 Winning % for Players where WS is HIGH and PER is LOW 39.0 59.0% 12.0 0.108 53.7% 50.9% 10.4% 8.6% 1.6% 1.7% 13.9% 14.2% 107 103
2011/12 Winning % for Players where WS and PER are consistent 33.7 51.1% 15.1 0.103 52.0% 48.2% 10.2% 14.3% 1.7% 1.7% 13.4% 20.0% 104 105
2011/12 Winning % for Players where WS is LOW and PER is HIGH 24.7 37.5% 15.0 0.064 50.8% 46.7% 8.8% 18.1% 1.7% 1.4% 14.0% 22.6% 101 108
2011/12 Additional wins for WS over PER players 14.2
2010/11 Winning % for Players where WS is HIGH and PER is LOW 49.6 60.5% 12.1 0.111 55.4% 52.3% 10.2% 9.3% 1.4% 1.5% 13.2% 14.8% 111 105
2010/11 Winning % for Players where WS and PER are consistent 39.8 48.5% 14.6 0.099 55.4% 52.3% 10.4% 13.3% 1.6% 1.6% 13.1% 19.3% 107 108
2010/11 Winning % for Players where WS is LOW and PER is HIGH 31.7 38.7% 15.4 0.071 55.2% 52.1% 9.1% 17.8% 1.6% 1.6% 13.7% 22.8% 104 110
2010/11 Additional wins for WS over PER players 17.9
2009/10 Winning % for Players where WS is HIGH and PER is LOW 50.1 61.1% 12.3 0.112 55.5% 52.4% 10.3% 9.0% 1.4% 1.6% 12.9% 15.0% 111 106
2009/10 Winning % for Players where WS and PER are consistent 40.6 49.5% 14.9 0.102 54.1% 50.3% 10.3% 14.0% 1.5% 1.7% 13.4% 19.4% 108 108
2009/10 Winning % for Players where WS is LOW and PER is HIGH 30.7 37.4% 15.3 0.072 52.8% 48.5% 9.1% 16.4% 1.6% 1.3% 13.3% 22.7% 105 110
2009/10 Additional wins for WS over PER players 19.4
Aggregate 4 season results
Winning % for Players where WS is HIGH and PER is LOW 49.4 60.2% 12.2 0.111 54.9% 52.0% 10.3% 9.2% 1.5% 1.7% 13.5% 14.7% 110 105
Winning % for Players where WS and PER are consistent 40.9 49.9% 14.8 0.100 53.5% 50.0% 10.2% 13.9% 1.6% 1.7% 13.3% 19.6% 106 106
Winning % for Players where WS is LOW and PER is HIGH 30.8 37.5% 15.2 0.069 52.5% 48.7% 9.2% 17.6% 1.6% 1.5% 13.7% 22.7% 103 109
Additional wins for WS over PER players 18.6
The ways in which HIGH WS players are different from HIGH PER players 2.4% 3.3% 1.1% -8.4% -0.1% 0.2% -0.2% -8.0% 7 -4
What is the main takeaway from all this? For players where WS and PER are similar, their respective teams play at a 50% rate, their average PER is at the league average for PER, and the WS/48 value is 0.100, which predicts a team winning 50% of their games. This is exactly as you would expect, the results should equal 50%. The significant result for me is the obvious and dramatic difference between the results of high WS and high PER groups. For the last 4 seasons, players who rate much higher in WS than they do in PER, play on teams that average 18.6 additional wins per season. WS/48 more accurately measures the contributions individual players made to their teams wins than PER does.
What are the areas of the game that this specifically relates too? First and foremost is in SHOOTING % and USAGE RATE. High WS/Low PER players have a TS% of 2.4% higher than High PER/Low WS players. High WS/Low PER players have an eFG% of 3.3% higher than High PER/Low WS players. High WS/Low PER players have a Usage rate that is 8% lower than High PER/Low WS players. PER does not penalize and actually encourages low efficiency players to shoot more, which has a directly negative impact on wins.
Looking at the individual statistical categories there is almost no difference between High Per players and High WS players in steals, blocks and turnovers. This doesn't mean that these metrics are unimportant, just that WS and PER handle them in a similar manner. There is a huge difference in assists. Does this mean assists are bad? No what it means is that either 1.) PER is giving to great of a weight to assists, or more likely 2.) players with high assist totals tend to be players who have high usage and a low TS%. The reason I believe #2 is the reason is because 2 of the 3 best players in both systems are high assist guys, LeBron James, and Chris Paul. What separates them from their peers is that they shoot well. The final metric is rebounding. There isn't a significant difference between High PER and High WS players, only 1.1%. What it indicates, at least to me, is that WS is not over weighting rebounds, it is placing an appropriate emphasis on rebounding. It is just the better rebounders shoot a higher %.
The two remaining categories are ORtg and DRtg. I will not attempt to explain each, I will let Basketball Reference do that, but ORtg and DRtg are used in the calculation of WS. As I understand each, after reading the book by the developer of ORtg and DRtg, Dean Oliver, is that he takes each teams total points scored and allowed per 100 possessions, and determines the relative contribution of each player towards their teams actual performance, through the use of statistical metrics. So the player with the lowest DRtg in the league is not necessarily the best defender in the league, but rather he was the best defender on one of the very best defensive teams in the league. The player with the highest ORtg is not necessarily the best offensive player in the league, but rather the best offensive player on one of the best offensive teams in the league.
When I post my final draft rankings they will be heavily influenced by this data. I place great emphasis on projected WS and Ortg and DRtg. Obviously players with high WS and high PER are good players, and players lower in both are players to avoid. I also focus a great deal on players who rate highly in WS and poorly in PER, and players who rate poorly in WS and very well in PER. Good WS/Poor PER players are players that often times are greatly undervalued, so their contribution cost is often much lower. Two players that jump out to me in this regard are Richard Howell and Andre Roberson. The players that have higher PER and WS where weighted to PER are players tro be concerned about. Among the players where the deviation is the greatest by general postion in the top 25;
Worst PG
Trey Burke
Best PG
Michael Carter-Williams
Best SG
Victor Oladipo
Worst SG
Archie Goodwin
Worst Forward Type
James McAdoo
Best Forward Type
Kelly Olynyk and Otto Porter
Worst Center Type
Cody Zeller
Best Center Type
Steven Adams



X
Log In Sign Up

forgot?
Log In Sign Up

Please choose a new SB Nation username and password

As part of the new SB Nation launch, prior users will need to choose a permanent username, along with a new password.

Your username will be used to login to SB Nation going forward.

I already have a Vox Media account!

Verify Vox Media account

Please login to your Vox Media account. This account will be linked to your previously existing Eater account.

Please choose a new SB Nation username and password

As part of the new SB Nation launch, prior MT authors will need to choose a new username and password.

Your username will be used to login to SB Nation going forward.

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

By becoming a registered user, you are also agreeing to our Terms and confirming that you have read our Privacy Policy.

Join Blazer's Edge

You must be a member of Blazer's Edge to participate.

We have our own Community Guidelines at Blazer's Edge. You should read them.

Join Blazer's Edge

You must be a member of Blazer's Edge to participate.

We have our own Community Guidelines at Blazer's Edge. You should read them.

Spinner.vc97ec6e

Authenticating

Great!

Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.

tracking_pixel_9347_tracker