Stats: most useful and most useless. (And various other complaints)

[nickagneta]

I just saw Roy answer apost that said Bynum was worth every penny of $80 million because his WP48 stat was so much better than Kobe's, so, LA should give Bynum the big bucks. I have to admit, I'm a mathematician at heart and love stats but even I had trouble coming up how this stat is relevant. Here is a site explaining how the stat is computed:

http://www.wagesofwins.com/CalculatingWinsProduced.html

I'm sorry, I find this a truly ridiculous computation in measuring the value or contributions of a player. And that's not the only one.

I hate the Per minutes stats. No one plays 48 minutes a game. Players that play only 20 minutes a game are only playing 20 minutes a game because they aren't good enough to play 40 minutes a game. Players who play 20 minutes a game and know that's all they are going to play can extend so much more effort and energy in that time span because they know they won't be playing bigger minutes. I find the entire idea to be an illogical concept.

Maybe I'm just an old fuddy duddy but give me points, rebounds, assists, steals, and shooting percentages every time and let me make my observations off of those and by what my eye sees. eFG%, PER, Per 48, WP, WP48, and some of these other stats just clutter things up if you ask me.

But then again so do just regular stats. I see so many people who measure defensive capabilities on blocks and/or steals alone which could be about the most misleading way of judging a defensive player as there can be. And in all honesty, I actually understand the concept of eFG% and how it relates. It's not a horrible stat and can give a better idea of a long range shooter's effectiveness, but I think it can also overexaggerate the worth of such players. Just my opinion.

Anyway, what's your most hate stat or the one you think is most useless. What is the most over rated stat. Interesting to know what some see as useful and useless.

[Roy Hobbs]

I'm right there with you on WP48, nick (as you know).  Any stat that says Andris Biedrins is a significantly better player than Paul Pierce is not a stat I pay much attention to.

As for "per minute" stats, I think they serve a purpose, but are often abused.  "Per minute" stats are irrelevant in terms of extrapolation; a player who produces a certain amount in 8 minutes is *not* likely to produce five times that level in 40 minutes.  However, I think they are fairly useful in comparing players of fairly equal numbers, to put them on a level playing field.  For instance, if you want a quick and easy way to compare one player who played 30 mpg, and another who played 34 mpg, looking at their "per 36" or "per 48" numbers are a fairly decent way to determine who had better stats.

I'm not really a fan of "+/-", because I think these stats are overly reliant upon: 1) the teammates a player is playing with, and 2) game situation (ie, do the bulk of minutes come against starters or in garbage time).  I know some members think that this stat proved that Rajon Rondo is the best player in the history of the universe, but to me, it's not a great indicator taken out of context.

It's the same thing with Opponent's PER:  this stat alone isn't a great indicator of a player's individual defense, because players who play in an up tempo system are going to get killed by it.  If an opposing small forward is constantly scoring points against your team on the fast break, does that mean that your small forward is a horrible defender?  No -- it means that your team gives up a lot of easy hoops due to its system.

One new age stat I really like is eFG%.  A lot of players have a lower overall FG% because they take a lot of three pointers; this levels that playing field, and shows who the truly great shooters are. 

I of course use all of these stats from time to time (except the completely and utterly useless WP48).  However, it's important to note that you have to use them in context, rather than manipulating them to win an argument.

[moskqq]

Another stat that haunts PG's is assists/game.  On teams that lack good shooters a PG is often frustrated because the ball is being delivered on time and in good position but the shooter fails to score. The low assist stat says that the PG is NOT effective but the reality is that the team has other glaring offensive deficiencies.

Another stat subject to misinterpretation is blocks/game and its relevance to team importance.  The Greg Oden's of this world are coveted BECAUSE of their shot-blocking capability.  Left unsaid is their ability to CHANGE shots or to force a perimeter game instead of a more reliable INSIDE game. Although no STAT tells the story the ability to change shots or to force a perimeter game certainly reduces an opponent's offensive efficiency.

There are certainly many more examples of the shortcomings of stats as a sole basis for measuring player worth.

[Roy Hobbs]

Another stat subject to misinterpretation is blocks/game and its relevance to team importance.  The Greg Oden's of this world are coveted BECAUSE of their shot-blocking capability.  Left unsaid is their ability to CHANGE shots or to force a perimeter game instead of a more reliable INSIDE game. Although no STAT tells the story the ability to change shots or to force a perimeter game certainly reduces an opponent's offensive efficiency.

Inversely, blocked shots can be misleading, in that a lot of blocks doesn't mean somebody is a good defender.  For instance, look at Sean Williams with B.C.  Somebody did an analysis that showed that B.C. was actually a *worse* defensive team with Williams on the court, despite his reputation.  Why?  Because he would often get out of position hoping to make a block, rather than making the more sound defensive play.

That's one of the reasons that I cringe when people use blocked shots to show that Darius Miles is a good defender.  From personal observation, Miles was a pretty poor defender, and average at his best.  However, his blocked shots paint a different -- and inaccurate -- story.

[wdleehi]

Another stat subject to misinterpretation is blocks/game and its relevance to team importance.  The Greg Oden's of this world are coveted BECAUSE of their shot-blocking capability.  Left unsaid is their ability to CHANGE shots or to force a perimeter game instead of a more reliable INSIDE game. Although no STAT tells the story the ability to change shots or to force a perimeter game certainly reduces an opponent's offensive efficiency.

Inversely, blocked shots can be misleading, in that a lot of blocks doesn't mean somebody is a good defender.  For instance, look at Sean Williams with B.C.  Somebody did an analysis that showed that B.C. was actually a *worse* defensive team with Williams on the court, despite his reputation.  Why?  Because he would often get out of position hoping to make a block, rather than making the more sound defensive play.

That's one of the reasons that I cringe when people use blocked shots to show that Darius Miles is a good defender.  From personal observation, Miles was a pretty poor defender, and average at his best.  However, his blocked shots paint a different -- and inaccurate -- story.

Right along with steals. 

I have seen fans talk about how AI is a good defender because of his steal numbers.  But if you watch him, you see that he plays the passing lanes, not the players.  And for every steal he gets there are 4 to 5 times that he doesn't get the steal, is out of position and the rest of the defense has to scramble to cover him.  Leads to easy shots.



Assists are misleading.  Many players with a high assist number is just because the have the ball in the hands all the time.  Either the shoot or make one pass where the other player has almost no option but to shoot. 





I think an underused stat is the Per compared to the defensive Per. 

[BudweiserCeltic]

Ah, this reminds me of the discussion this offseason about the usage of Turnover Percentage as a stat to prove that Ginobili is not a turnover machine.

[guava_wrench]

I'm not really a fan of "+/-", because I think these stats are overly reliant upon: 1) the teammates a player is playing with, and 2) game situation (ie, do the bulk of minutes come against starters or in garbage time).  I know some members think that this stat proved that Rajon Rondo is the best player in the history of the universe, but to me, it's not a great indicator taken out of context.

That's why the RBI stat in baseball is dumb. It measures opportunities as much as success.

[JSD]

Hands down, a pitchers Win-loss record.

[drza44]

I think the key is that there is no "holy grail" stat that is perfect for every situation.  Stats need to be used correctly, in context, to make specific points within an argument (not to be the whole argument).  

On the other hand, some people go too far the other way in suggesting that stats are useless.  I think that, if used correctly, most of these stats are extremely important to really gauging how a player compares with his contemporaries.

As far as stats that I like, many of them have been mentioned here.  I'm more of a fan of the APBR-metric stats than the traditional points/rebs/assists because I think it is harder to pad APBR-metric stats than it is to pad something like points.  I like the wins produced stat and the Roland Rating (82games.com, uses +/-, PER, and opponent PER) in general because both at least try to take into account offense and defense.  I don't love steals or blocks, for the reasons stated above, as many use them to make arguments on how good a player is on defense and I don't think steals/blocks correlate very well with defensive effectiveness.  To that end, I don't love PER because it is used as an "overall player" rater when in fact it doesn't do much at all to measure defense.

But to me, the best arguments can find support across a wide variety of stats.  If wins produced, Roland rating, and PER all tell a similar story then I give it more weight than if every stat disagrees with every other.

[carlherrera]

I agree with most posters here: you need to look at more than just one stat to get a good picture of a players' performance. That's why there is WP and WP48 so if for instance, Bynum or Biedrins can't stay on the floor because of foul trouble, you will see that their WP are low while their WP48 are high. Showing only the WP48 can be misleading - although still very valuable. You still need to look at stats that try to project a player's full performance though - including both defense and offense - like WP does.

Having said that, Roy, WP is a PROVEN methodology, while not perfect, time and time again it correlates incredibly high with that player's team's wins - i.e. a team that won only 25 games has players with WP stats that add up to really close to 25 wins.

So there are two possibilities here: A proven system is wrong or you are right just because you say you are right. I think the answer is somewhere in the middle - although more towards the WP middle WP does not explain everything but it does a great job of explaining a player's overall performance, specially when combining it with WP48. You have a great knowledge of the game but I can't just take your word for it. I need evidence!

Finally, I think basketball is experiencing what baseball experienced in the 90s with Sabermetrics. There are new stats that are better but the status quo won't accept them because it goes against what they have believed for many years. Teams that take advantage of this transition will benefit greatly just like the Oakland A's benefited in the last few years until other teams caught up. (See Red Sox and Bill James).

[nickagneta]

I'm sorry I just can't give to much credence to a stat that slants the stats based on the position played and bases a good portion of the statistic for an individual on the performance of the team as a whole like WP does.

Weighing the stat in favor of position relative to the league average and how the talent performs at each position in a given year is wrong when trying to measure the value of an individual. Each team has a different dynamic with each position having different difficulty to play and different importance. PG within the Hornets is different than PG within the Celtics. PG is not nearly as important a position in the Celtics as PG is for the Hornets or SF is for the Celtics. Within a triangle offense certain positions are easier to play and more or less important than Mike D'Antoni's offensive system. Yet the players are measure based on a weighted system that all PG position are harder to play than center positions and that everything is judged by a league average.

Also team performance is then icorporated into it as well. You could be the best player in the league but because of poor talent and overall play around you have a low WP and the opposite also holds true. Take a look at the top players for 2006-2007: 

1 Jason Kidd 24.8
2 Kevin Garnett 20.6
3 Dwight Howard 20.5
4 Tim Duncan 20.1
5 Shawn Marion 19.5
6 Steve Nash 19.4
7 Carlos Boozer 18.4
8 Dirk Nowitzki 18.0
9 Marcus Camby 17.6
10 LeBron James 17.4
11 Tyson Chandler 16.9
12 Ben Wallace 15.8
13 Kobe Bryant 15.3
14 Luol Deng 14.9
15 Manu Ginobili 14.1
16 Emeka Okafor 14.1
17 Elton Brand 13.6
18 David Lee 13.6
19 Amare Stoudemire 13.4
20 Chauncey Billups 13.3
21 Chris Paul 13.2
22 Chris Bosh 12.6
23 Vince Carter 12.5
24 Andre Iguodala 12.4
25 Al Jefferson 12.2

Am I to believe that David Lee was such an outstanding player in 2006-2007 that he played better than Chris Paul, Chris Bosh, or Andre Iguodala? Am I to believe that LeBron had a worse year than Marcus Camby or that there were 12 players in the league that were better than Kobe Bryant that year?

I'm sorry, I think this stat is seriously bogus. Spend some time reading where they contrived the numbers from. There is a lot of conclusions and variables that are used that have nothing to do with the players individual performance.

[Sweet17]

Nick is right about WP. He correctly says why it's broken. Simply put teams have different roles for different players in their schemes and thus you can't judge players without taking this into account.

To flesh this out a guy like David Lee doesn't create any shots on his own. He doesn't need to direct an offense or handle the basketball. He just concentrates on rebounding and finishing with dunks. Thus he rarely screws up. This makes him a god on this win produced statistic. But SOMEONE has to do that hard work on that team.. This stat takes no account of the roles a guy might be asked to do..

When it comes down to it right now individual judgement is STILL a better way to judge talent.

There is absolutely no chance that David Lee is better then Chris Paul. That stat fails the laugh test... 99% of the GMS around the league would take Iggy over David Lee. Iggy does everything you need on the basketball court. It's a no brainer..

Pete

[carlherrera]

I agree that WP is not perfect, specially when comparing players at different positions.
However, it is at the very least a good tool to understand a player's value.
If it were so easy to just concentrate on rebounds and easy put backs, how come noone did it better than David Lee in 06-07? Becuase he was better than others doing it, that's why.
If Kobe was so great in 06-07, and he had to handle the ball so much, create shots for his teammates and take difficult shots, how come he didn't have a better effiency, more assists, etc than other similar players? Because he wasn't that good, that's why.
We have been programmed to look at the Sports Center highlights and listen to many "experts" that are no experts at all and believe that Kobe is the best simply because that's what we like to believe.
I'd rather look at MULTIPLE evidence and make up my own mind. I don't think WP should be the ONLY data point, but it is a one that must be considered.
Re: Chris Paul, of course he's amazing, he made a huge improvement last year and WP corroborates it!

[brownbagger]

Im in line with what the OP of this thread is saying but when someone is in need of making a point

on CB or debating to the end usually it never fails we get a bunch of stats thrown our way to make

a point.Over 75% of the time the stats thrown out there are way off base in making a point.I find

some people can't accept others point and move on,they have to google or come up with endless stats.

What i  see before me will always win out over stats of players anyday,usally there are circumstance

as some have pointed out to accompany them.

[Greenback]

“In baseball, my theory is to strive for consistency, not to worry about the numbers. If you dwell on statistics you get shortsighted, if you aim for consistency, the numbers will be there at the end.”   ~Tom Seaver