Friday, December 21, 2012

Traditional vs. scoring point guards: a statistical analysis


The NBA of today looks quite different from the game that was played a decade or two ago. With the disappearance of the center from the 2013 All Star ballot, no longer do scouts and general managers seek the monstrous talents like Shaquille O’Neal and Tim Duncan but prefer forwards like LeBron James and Kevin Durant who are multi-faceted players (or at least strive to be). Specialization in one skill is no longer coveted by coaches and players looking for the next great thing. Players are expected to be well-rounded—possessing size, speed, and basketball skill. Point guards like Ricky Rubio are flashy but youngsters like Kyrie Irving and Russell Westbrook are typically regarded as more electrifying. Unlike Rubio, those players gifted scorers. This attribute is sought after because when a play breaks down, they do not always have to look to teammates to shoot, but rather have the option to create—and make—their own shot.

So, I thought that it would fruitful to look at the statistical successes of the current crop of point guards. Ultimately, the goal is to see if it is possible to ascertain which type of point guard is better: the scoring one (which is gaining popularity), or the traditional one who is a pass-first play-maker.

Here’s the general idea: I have created a statistic that gives players a traditional-to-scorer rating (TSR) and then will compare that rating to the success of their team. For the purposes of this analysis, I have chosen winning percentage and whether or not they make the playoffs to be that measure of success. Undoubtedly, a point guard like Russell Westbrook has more to work with than Ramon Sessions, but the idea is that by using a large sample size (namely the NBA as a whole), the outliers will have less of an impact considering they are only one of 33 point guards.

TSR uses field goal percentage (FG%), field goal attempts per 48 minutes (FGA48), assists per 48 minutes (AST48), and assist-to-turnover ratio (AST/TO). A scoring point guard would have a relatively low FG% because he is taking more shots, a higher FGA48/AST48 ratio, and a higher TO/AST ratio (which is the inverse of AST/TO ratio). I used per 48 minute statistics to even out the players statistics that may have been inflated or deflated based on minutes played. I believe this to be the most efficient way to measure a player’s true ability because all of these guards play generally around 30 minutes a game, but some closer to 35 and others closer to 25. Anyone who played less than 25 minutes per game (essentially half of a four-quarter NBA game) was not included in this analysis.

Conversely, a traditional point guard likely has a higher FG% because he takes more high percentage shots, a lower FGA48/AST48 ratio, and a lower TO/AST ratio. Without getting into too much more detail (as if this isn’t enough already), the final formula for raw TSR is as follows:

[(FG%)-2 x (FGA48/AST48)] + TO/AST

In order to make sense of this number, I will compare it to the median point guard. Thus, the final TSR rating = raw TSR – median PG TSR. This gives the TSR of the median point guard a value of 0. The more of a scoring PG the player is, the more positive the TSR; the more of a traditional PG the player is, the more negative the value. Using TSR, here is a graphic (click to enlarge) of where six prominent scoring PGs and six prominent traditional PGs fit onto the spectrum:
TSR of 2012 point guards, as of 12/18/12.
    
Ramon Sessions 
6.9
Rodney Stuckey 
6.4
Brandon Jennings 
6.4
Raymond Felton 
4.2
George Hill 
4.1
Kyrie Irving 
3.8
Brandon Knight 
3.7
Alexey Shved 
3.3
Stephen Curry 
3.1
Kemba Walker 
3.0
Damian Lillard 
2.9
A.J. Price 
2.1
Kyle Lowry 
1.9
Russell Westbrook 
1.3
Jameer Nelson 
1.2
Deron Williams 
1.0
Ty Lawson 
0.0
Jeremy Lin 
-0.3
Jeff Teague 
-0.5
Luke Ridnour 
-0.6
Kirk Hinrich 
-1.3
Jrue Holiday 
-1.6
Darren Collison 
-1.7
Mike Conley 
-1.8
Mo Williams 
-2.0
Goran Dragic 
-2.2
Jarrett Jack 
-2.5
Tony Parker 
-2.7
Greivis Vasquez 
-2.8
Jason Kidd 
-2.9
Jose Calderon 
-4.7
Chris Paul 
-5.1
Rajon Rondo 
-7.3
Now, to answer the question: which point guard is better for his team?

To do this, let’s look at the player’s TSR and compare that to winning percentage and playoff seeding. Of course, this will certainly change as the season progresses, but as of mid-December, the 2012-13 point guards show an interesting trend. Again, a traditional point guard has a negative TSR (see table to the left, all statistics as of December 18) and a scoring point guard has a positive TSR.

Ty Lawson happened to be the lucky man in the middle and is thus excluded by convention of TSR calculation. Now for the results:

        ·         TSR shows that 7 of 16, or 44 percent, of scoring point guards have a winning percentage of at least 50 percent (which is the “average,” by default). Additionally, 44 percent of scoring point guards are playoff-bound (according to their seeding).

How about those traditional point guards?

        ·         TSR shows that 11 out of 16, or 69 percent, of these point guards have at won at least half of their games; 10 out of 16, or 63 percent, of traditional PGs are on playoff teams.

Based on TSR, 2012-13 PGs are more successful if they are traditional (pass-first) as opposed to scorers.

The next logical question would be: what about comparing only the players who did make the playoffs? After all, teams that struggle may look to their point guard to be more of a scorer (see Ramon Sessions, Rodney Stuckey, and Kyrie Irving).

Just looking at the above statistics, there were 17 point guards out of 32 that would make the playoffs if the NBA season ended in mid-December. Out of those 17 players on playoff teams, 10 of them, or 59 percent, are traditional PGs.

TSR thus seems to at least lean us to show that traditional point guards are more successful.

Of course, it could not hurt to see how this conclusion stacks up against previous seasons. So, I did the same quantifications for the 2011 point guards (see the table below). Statistics for these PGs based on TSR:

·         53 percent of scoring point guards had a winning percentage above 50 (8 out of 15)
                                                                    ·         73 percent of traditional point guards had a winning percentage above 50 (11 out of 15)
Shannon Brown
38.8
Randy Foye
17.6
Iman Shumpert
10.4
Kemba Walker
9.7
Avery Bradley
8.6
Brandon Knight
7.7
Brandon Jennings
7.5
Russell Westbrook
6.8
Monta Ellis
6.0
Tyreke Evans
5.5
Jrue Holiday
5.4
Deron Williams
2.0
Mario Chalmers
1.3
Isaiah Thomas
0.6
Jameer Nelson
0.2
Raymond Felton
-0.1
John Wall
-0.7
Darren Collison
-0.7
Jeff Teague
-0.9
Mike Conley
-1.2
Devin Harris
-1.5
Tony Parker
-1.9
Goran Dragic
-2.1
Ty Lawson
-2.2
Greivis Vasquez
-2.3
Chris Paul
-3.2
Andre Miller
-3.7
Rajon Rondo
-6.7
Jose Calderon
-6.9
Steve Nash
-7.3
       ·         40 percent of scoring PGs were on playoff teams (6 of 15)
       ·         60 percent of traditional PGs were on playoff teams (9 of 15)
       ·         40 percent of playoff PGs were scorers (6 of 15)
       ·         60 percent of playoff PGs were traditional (9 of 15)

         Of course, it would be inaccurate to surmise that this implies that traditional PGs are better than scoring PGs.
   
Russell Westbrook, Deron Williams, and the six scoring PGs mentioned in the graphic are better than the majority of PGs in the league. But it is worth nothing that the collection of players who are at the top of the league follow a similar pattern as to that of the league as a whole. This pattern shows that traditional point guards generally have more success.

A qualitative comparison along with some hypothetical questions also sheds some light on this debate. Take the Lakers, for example: a star-studded team with one of the NBA’s greatest players leading the way at shooting guard. Would Russell Westbrook work at point guard? Would even someone like Deron Williams or Stephen Curry work?

The answer to Westbrook is a resounding ‘no,’ and the other two are up for debate. What is more obvious is the fact that choosing a PG from the opposite end of the TSR spectrum would likely work better. Rajon Rondo, Chris Paul, Tony Parker, or Jrue Holliday all fit the bill better for what is going on in Los Angeles. Thus, the Lakers decided to pick up an aging Steve Nash who is the ultimate traditional PG (according to TSR and popular opinion).

Los Angeles is an extreme example because only a handful of other teams have rosters as star studded as the Lakers. They are were a model of consistency for over a decade. But a look at a team with fewer household names gives us the same result.

Golden State traded Monta Ellis last year for Andrew Bogut and a host of other players that have had a marginal impact this season (Richard Jefferson, etc.). In that regards, the Warriors essentially lost one of the best scoring PGs in the league for nothing. While it is questionable to list Monta as a PG—I’m taking ESPN’s word on that one. Every statistic that I pulled up listed him as a one guard, not a two. And for the amount of ball handling he did, that is a reasonable conclusion. Taking Monta as a PG and completely removing him from the equation in Golden State, what do we have as a before and after?

Monta broke out in the 2009-10 season, averaging 25.5 PPG and gaining national attention as one of the best scorers in the league. From that year through last year (and the trade to Milwaukee) the Warriors never won more than 44 percent of their games. Stephen Curry was on the team, David Lee and Dorell Wright were there, even coach Mark Jackson and then-rookie Klay Thompson tagged along for the latter part of that three year stretch.

But the Warriors were awful—as a team. Monta's talent and a dedicated fan base packed Oracle on a nightly basis, hoping to bring back some of that magic that enveloped the “We Believe” 2007 playoff run. Until this year, the only thing worth watching was the incredible scoring ability of Monta (who averaged 26, 24, and 22 PPG for that three year period).

A look at the re-inspired Warriors of 2012 says everything. They have begun this season with their best start, 17-9, in years and a 6-1 road trip that included a win over the Miami Heat. That trip rivals road performances by some of the best teams in NBA history (Spurs of last year and Lakers of early 00s, among others). Warriors GM Bob Myers was quick to point out that it is only the first quarter of the season, but there is no denying that the culture within Golden State has changed. The previous culture created by the scoring nature of their former (point) guard, Monta, was transformed into a team-oriented style where Stephen Curry and Jarrett Jack are able to play off of each other’s strengths as scorers and facilitators. More importantly, their scoring does not sacrifice team chemistry.

Curry isn’t exactly a traditional point guard with a TSR of 3.1, but Jack is more of a facilitator (TSR of -2.5). Both are certainly looking for their shot but they are willing facilitators. And they are light years more "traditional" (or more willing to pass-first) than Monta Ellis (TSR 6.0).

This is where I believe TSR is an excellent statistic. We can see that Curry is a scoring guard but he leans more toward the median than Monta Ellis, and still further toward the Nash/Rondo side is where we find Jarrett Jack. It gives us a quantifiable measure of a PGs tendency to be a pass-first or shoot-first player and then we can compare them as such.

The conclusions from my brief study of the past two years point guards brings to light an interesting thought, and one that I still believe to be true despite the popularity of players like Derrick Rose and Kyrie Irving. Traditional point guards should still be valued in the NBA today. TSR puts some weight behind that thought.

One final remark: I have been a critic of Westbrook’s style of play, especially as it pertains to three-time scoring champion Kevin Durant. KD, in particular, would be better off with a traditional PG...in my opinion. In my pre-season ranking of the top 10 players in the NBA, I said that Westbrook could be a top-five player with “some modification to his gameplay.”

OKC has won 12 out of their last 13 games. As for modification to his style of play, let the numbers speak: Westbrook’s TSR last season was 6.8 and this season it has dropped drastically to 1.3.

-------------

If you have read this far, let me know what you think of this new stat, TSR. Do you think it uses relevant statistics to accurately predict a PGs “traditional-to-scorer rating”? If you have some thoughts as to the result of my brief study, please feel free to elaborate (and ultimately that was the goal of creating TSR), but any thoughts on the statistic itself are particularly appreciated. If you need any clarification or would like a look at the data analysis, you can email me here.

TSR, or “traditional-to-scorer rating,” is an original statistic that was created by the author of this article and its full rights belong solely to this website, Bases and Baskets.

4 comments:

  1. Very good work man. You've always loved Rondo more than Westbrook, and Curry more than Ellis, and now you have a stat to back it all up!!! Haha. Good work! I think that the stat is very useful, especially with the recent advent of the scoring PG, and your analysis does show that scoring guards, as a whole, win less than traditional ones. Now, for some constructive criticism: (Note: I will use raw ast and fga, because idk where the 48 min are, but I am mostly comparing players who play similar minutes)

    On the formula itself: I mentioned this in a message earlier, but I would suggest weighing FG% less, or equal to, the FGA48/AST48 ratio. You are right in that a scoring guard takes more shots, which largely results in a lower FG%, while a traditional guard takes fewer shots but shoots a higher percentage. This is true in general, but there are a lot of other variables. The primary one is the situation in which the player is in and the shots the player takes. For example, Deron Williams has long been a similar player to CP3, but his TSR is significantly higher although their assists per game (9.2 vs 7.7) and FGA (13.1 vs 11.7) are quite similar. The large discrepancy is in FG%, where CP3 is at 47%, while D-will is at a putrid 39%. This does highlight differences in their games - Dwill takes lots more 3s than CP3, and he's also a less effective player - but I wouldn't quite call Dwill FAR more of a scoring PG than CP3 is, which is what TSR suggests.

    Another example is Tony Parker. Long known as a scoring guard in an age of traditionals (mid-2000s), TSR projects him as a traditional guard, significantly more traditional than Deron Williams, although Parker has fewer assists (7.2 vs 7.7) and more FGA (15 vs 13), so he is certainly more of a scoring guard. The discrepancy occurs primarily in FG%, where Parker shoots >51%! This is because Parker is a slasher who gets to the rim for high percentage shots, while Dwill shoots a lot of jumpers/3s. I feel that your formula could benefit by weighing FG% as much, or less, than the ratio, or possibly including some other stat (eFG%, for example) that takes into account differences between Parker and Williams shot selection.

    Next, I can do a brief discussion about your analysis as a whole. While you make a good point overall, remember, you don't get assists on missed baskets. This may be a reason why some players on bad teams get labeled as scoring guards (and thus, overall scoring guards are less successful than traditional guards) - they actually do pass, they just don't have teammates that make shots. Would Rondo average 12 APG playing on the Charlotte Bobcats, in place of Kemba Walker, with Bismack Biyombo and Tyrus Thomas? Probably not. I'm not saying your point is completely false, but you should consider that it is easier to get labeled a scoring guard if you play on a bad team because you will generally get fewer assists. Some players that came to mind include Kemba, John Wall, Rodney Stuckey, and Jameer Nelson (in his case, I believe he would be more traditional in '10/11 when D12 was on the team).

    Either way, this is a very good stat, and I feel that it will be even more insightful with a few tweaks. Good job!

    ReplyDelete
  2. Haha thanks and yeah it does lean toward favoring Rondo and Curry. Don’t forget the very last point of the article though! Westbrook has changed and OKC has been on a roll.

    One of the main reasons I decided to weigh FG% more was to create a continuum with a larger ranger that was easier to grasp one’s head around rather than having the numbers be .001 vs .002 (slight exaggeration, of course, but you get the idea). As for your insight into the D-Will/CP3/Parker comment, one quick cautionary note is that these numbers are specific to the
    past two years—not the mid-2000s.

    But with that in mind, weighing FG% less and including a stat like eFG% is plausible and would lean D-Will further left in the TSR spectrum. If FG% alone were to be decreased in weight in the TSR calculation, it would still label D-Will as a scoring PG. But it is definitely worth experimenting with further because eFG% does help neutralize differences in shot selection. If I do write a follow-up piece, your suggestions will certainly be taken into account. Good thoughts.

    You’re final point is true and probably the thing that TSR does struggle to account for. PGs on bad teams are probably forced to score more than they would if they were on a better team…but the counter of that is Russell Westbrook. He has every reason to be a traditional PG and still manages to have—this very season—more shots than Kevin Durant. This pinpoints the
    problem I have with Westbrook and why I like Rondo. Rondo does what he is supposed to as a PG, he facilitates for others before creating for himself. And unlike Westbrook, he doesn’t have someone like KD to dish to. I do agree that Rondo probably wouldn’t average 12 APG on Charlotte, but I still believe he would lean well to the “traditional” side of the spectrum because he isn’t a consistently capable scorer yet (maybe he would be if given the opportunity).

    In my opinion, a point guard is meant to be the field general. Team chemistry starts with him. Some teams have point-forwards (LeBron) which fulfill similar roles, but for the most part, a point guard’s job is to dish. Maybe I’m “traditional” (pun intended) in that thinking but like you said, this stat does give that notion some support.

    ReplyDelete
  3. We're very specialized in what we do in the variable annuity business, but that's that's the nature of social care in a way that continues to yield new benefits for patients, employers, and care providers. And suddenly the world seemed a little less liquidity than a public corporate, and the coverage amount.

    Feel free to visit my web page ... Best Car Insurance

    ReplyDelete

Add your thoughts or opinion.