20 Second Timeout is the place to find the best analysis and commentary about the NBA.

Thursday, November 06, 2008

Where Does Tony Parker Rightfully Rank Among NBA Point Guards?

Tony Parker's 55 points and 10 assists in a 129-125 double overtime San Antonio victory over Minnesota are significant for reasons beyond the fact that Michael Jordan and Oscar Robertson are the only other players in NBA history to reach those point and assist totals in the same game.

There are three ways to look at what Parker did:

1) This is a fluke performance.
2) Parker's skill set remains the same but the situation necessitated that he assume a bigger offensive role on the team.
3) This is a breakout game signifying a quantum leap in the quality of his skill set.

These three statements are not entirely mutually exclusive. Clearly, considering the historical context that only two players--two all-time greats--have ever matched these numbers, it is highly likely that Parker will never have another game quite like this. In that sense, the numbers are fluky (that is not at all the same thing as saying that Parker merely got "lucky" to play so well). It is obvious that with Manu Ginobili out of the lineup both Parker and Tim Duncan are shooting the ball much more frequently than usual. That said, perhaps the coming weeks and months will confirm that Parker has taken his game to another level; he does not have to put up 55-10 every night to prove that but if he suddenly becomes a 25-8 player for an extended period then that would be a big improvement over his career averages of 16.1-5.5 and his single season bests of 18.9 ppg (2006) and 6.1 apg (2005).

While no one could have predicted that Parker would have this kind of single game output, considering his skill set--blazing speed, excellent ball handling abilities, good finisher in the paint, erratic but sometimes deadly jump shot--it makes sense that he can score 50-plus points when everything comes together: high number of field goal attempts (36, five more than his previous career-high), excellent field goal percentage (.611, significantly better than his already good career norm of .488), better free throw percentage than usual (9-10) and a couple three pointers thrown in for good measure (Parker only made 17 three pointers in the entire 2008 season).

When teams were lining up this summer to offer big money deals to Gilbert Arenas and Baron Davis, I did a post that ranked the top point guards in the NBA based on their skill sets: Chris Paul finished first, followed by Steve Nash, Deron Williams and Tony Parker. In other words, long before Parker made the highlight shows on Wednesday night, I considered him to be an elite point guard, a better player than the more heralded Arenas and Davis.

All things considered, Parker's outstanding performance does not really alter my opinion of his game--at least not yet. The reason that I don't have to alter my opinion is that my opinion was based on a skill set evaluation of his game, as opposed to blindly looking at numbers or being mesmerized by a player's "swag" and other irrelevant considerations. San Antonio's system normally has a suppressing effect on individual player numbers because Gregg Popovich's limits the minutes of his star players and those three stars split up the offensive duties pretty evenly. Based purely on their skill sets, each of those players could put up bigger numbers, so it is not correct to evaluate them solely on their statistics.

Parker is a top notch point guard who had a great performance; the absence of Ginobili and the fact that the game went to double overtime further contributed to the numbers that Parker put up.

It is interesting to look at one "stats guru's" take on all of this. According to John Hollinger's "adjusted game score" calculations, Parker's performance is the 43rd best single game performance in the NBA since the 2001-02 season. Amare Stoudemire's 49 point, 11 rebound, six assist game versus Indiana last night ranks 15th on Hollinger's list, largely because Stoudemire shot a much better percentage than Parker and Stoudemire did all of his work in regulation, though Parker actually only played seven more minutes than Stoudemire did. By the way, Hollinger asserts that Kobe Bryant's 81 point game in 2006 is easily the best single-game performance of the past seven years; Bryant has three of the top four games and five of the top 14.

Hollinger ranked Parker as the top player in the NBA so far this season even before this game; now he credits Parker with a stratospheric 39.41 Player Efficiency Rating (PER). To put that in perspective, according to BasketballReference.com, Michael Jordan is the career PER leader with 27.91, while the best single season PER ever is Wilt Chamberlain's 31.84 in 1962-63. Granted, Parker's rating is only based on four games--and he played at a historically great level in one of them--so this is a bit like a baseball player who hits .400 in the first month of the season before settling back to his regular level but I still think that this is a good example of how numbers crunchers often miss the big picture: in 2007-08, Parker did not even make the top 20 in Hollinger's PER rankings but now we are supposed to believe that he is playing at a higher level than Michael Jordan and Wilt Chamberlain did? Parker's raw box score numbers--the figures that "stat gurus" deride as meaningless--are 33.3 ppg, 7.3 apg and .564 field goal shooting through the first four games. Yes, I realize that once there is a larger sample size for Parker's games his PER number will regress to a more sensible level but my point is that this example gives a vivid demonstration of how PER does not accurately and meaningfully quantify a player's abilities, either over a four game stretch or a season--unless you think that Parker is playing significantly better than Michael Jordan and Wilt Chamberlain did when they were at their best.

Labels: , ,

posted by David Friedman @ 4:33 PM

12 comments

links to this post

12 Comments:

At Thursday, November 06, 2008 5:38:00 PM, Blogger Steve said...

I've always felt that Tony Parker has never received the attention he deserved. The guy was the 2007 Finals MVP for crying out loud.

I think that I lot of people are going to dismiss this performance simply because it came in a 2-OT game against the Wolves, and also because of the performances of Amare and Lebron on the same night.

I however think it was great. That jumpshot to send the game into 2-OT was fantastic. Parker has really improved his jumper the past few seasons. One of the knocks I have on him is that he's never been much of an assists guy. However the fact that he's managed to average some 18 ppg in about 32 mpg is remarkable.

If it takes some 85 combined points from Parker and Duncan to beat the Timberwolves in 2-OT, then the Spurs will be in for a long season.

 
At Thursday, November 06, 2008 6:28:00 PM, Anonymous Anonymous said...

Sort of a strange piece here. I don't know why Tony Parker on average couldn't have performed better, over four games, than Jordan did over his entire career on average. It's four games. Big deal. The stats aren't saying anything other than that Parker has had four great games. He might go out tomorrow, shoot 8-26, and all of sudden be back to normal. He probably will.

And I don't what you mean about stat gurus deriding raw stats as meaningless. Certainly they are meaningful, although when comparing one player to another per minute numbers are more useful.

I posted earlier on another thread. As I said there, stat gurus differ on how to divide credit at the individual level.

Hollinger, PER, and Game Score, wil give you one view. Berri, Win Score, and Wins Produced will give you a different view. However, everyone would agree that Parker had an unbelievable game. He posted a WS of 22 in 50.5 minutes.

My choice for best performance? Manu on Feb 13, 2008, which was the highest PawsMin in the last two years according to NBA Babble (which is no longer working unfortunately.) 46 points on 15-20 from the field, 8-11 from 3, 8-9 from the free throw line, 5 rebounds, 8 assists, 3 steals, a block, and just one turnover in 40 minutes of play. That's a WS of 31. I don't have numbers further back than that but Kobe was slightly more productive with a WS of 32 in the 81 point game.

i would say Tony Parker is a top ten guard but decidely in the second tier of NBA point guards behind Kidd, Nash, Paul, Billups, Calderon, and Deron.

You don't have all that much love for Calderon yet. I have a feeling you are going to say he gets a lot of credit that really belongs to all the amazing shooters they have assembled around him.

Owen

 
At Thursday, November 06, 2008 7:25:00 PM, Anonymous Allen said...

David,

I think you're misinterpreting the meaning of Tony Parker's PER score through four games.

The PER indicates that for four games Parker played better than Jordan and Chamberlain did on average for their careers.

This is not that implausible a statement. Think about taking four of any all-star players' best games in a season, and I'm sure that those games were better than MJ and Wilt's average games over their career. It's not saying much, because everyone has good games, and Parker just happens to have had four great games to start the season.

 
At Friday, November 07, 2008 5:47:00 AM, Blogger David Friedman said...

Allen:

I understand that this is a small sample size and I readily acknowledged that in the post--but look at Parker's boxscore numbers in those four games: 33.3 ppg, 7.3 apg, .564 fg%. How do those numbers add up to a rating that is not just equal to MJ and Wilt's career ratings but significantly better? Look at the PER numbers that I listed in the post. That just does not make sense and I think that this is an example of how PER and other stats that purport to make categorical ratings are missing something.

 
At Friday, November 07, 2008 6:21:00 AM, Blogger David Friedman said...

Owen:

Parker could have performed better than MJ's career over four games but do his raw boxscore numbers indicate that he did so, let alone by a significant margin? I just don't get how the numbers he put up in those four games compute to such a historically great PER--and that suggests to me that there is something wrong with PER, some inadequacy in the way that it crunches the numbers. Parker's first four games have been great but I would not take this Tony Parker over an average MJ performance or an average Wilt performance.

Perhaps I am mistaken, but my understanding is that--at least in some quarters--boxscore numbers are considered passe and are only useful as something to be plugged into a formula like PER. I agree that per minute data can be more useful than per game data in some situations--not all--and that raw fg% does not convey as much information as TS% or adjusted field goal percentage but I still think that there is value in box score data. The bottom line is wins and wins are defined by game to game performance. I know that there is a big debate about how to extrapolate per minute data of players who average 10-15 mpg in order to compare it to players who average 35-40 mpg. We have discussed that type of issue a bit when comparing Kobe to Ginobili, though the difference in their mpg is obviously not that great. I am definitely in the camp that believes that you cannot simply look at a player's per minute production and assume that he can maintain it with additional playing time; I would have to see a low mpg player play against good competition and evaluate his skill set to make that determination.

It does not surprise me that you think that Manu played the best game during the time period in question :)

I put Parker in the first tier of pgs, but behind Paul, Nash and Deron Williams.

I don't have anything against Calderon's game per se but I'm not ready to put him in with that top four. I have no problem putting him in that second group. Calderon shoots and passes really well. Is he capable of consistently scoring the way that Paul, Williams, Parker and Nash do? That has not been proven yet, though he is averaging close to 20 ppg so far this season. I don't think that Calderon is a great defender, either. I had DWill ahead of Paul until last season and then I moved Paul ahead of all of the pgs, so I am certainly open to moving Calderon up the charts if his performance warrants doing so. I know that Toronto is very high on him because they dealt Ford, who is a quality player.

 
At Friday, November 07, 2008 4:00:00 PM, Anonymous Anonymous said...

D'Antoni was qouted this morning in reference to Duhon's having scored just nine points, "Nash scores zero and is one of the best players in the league,” D’Antoni said, referring to Steve Nash, the guard who orchestrated the D’Antoni system in Phoenix. “Individual stats mean absolutely nothing.” I think D'Antoni is responding to the pervasive view of statistics, which is that raw scoring, without regard to efficiency, is what matters. He is basically making the case that you can't judge a player by his scoring totals.

Because with Duhon, you can definitely judge him by his individual stats, if you go deeper than scoring. Thus far, he has posted a WS of 7.4 per 48, slightly above average for an NBA point guard, due to low turnovers, good rebounding, decent passing, and 1.3 steals per 36.

I wouldn't say that box score data in general is passe. I would say what's passe amongst statheads are traditional abuses of box score data, like the idea of the importance of a "20-10 guy (Zach Randolph) or a 20-8 guy like Marbury. Marbury, as we have discussed has the largest contract in the NBA based off his 20-8. I would say that would be a case where I agree with D'Antoni that the stats mean nothing.

Re Parker and Per- Having looked at the numbers, I stand corrected. I agree with you.

As I said previously, there is very little disagreement about the value of team level statistics in the NBA, and a great deal of dispute over how to divide credit at the individual level. As a WOW guy, I would argue that PER dramatically overrates scoring and this would be a case of that. When I compute Parker's WS per 48 through the first 4 games at 13.26. Chris Paul last year was at 16.68 per 48 on average. And Jordan averaged something like 18.333 for his career, (a number which understates his real value, since it includes his last two seasons, and doesn't include the defensive team adjustment credit added into Wins Produced.)

I think the evaluation of Manu and Kobe's games really show you the difference between Hollinger and Berri. Game Score will tell you that Kobe's game was much much better than Manu's (46.6 to 64), while WS will tell you it was just marginally better (31-32). It comes down to a difference in how they approach the issue of fga's. The more shots you take, the higher your Game Score will be.

Re per minute and per game numbers, everything has its place, I agree. Carl Landry posted unbelievable numbers last year. Literally eye popping. And he also had an great adjusted +/- number too. Does this mean he is the best player on the Rockets? Well, no. But it certainly might mean that Orlando should have signed him rather than Pietrus, who they gave more than twice as much money.

Calderon is off to a great start. I can at least thank my knowledge of advanced statistical methods for getting me to pick him off the waiver wire in my keeper league. His statistical profile was pretty clear as soon as he came into the league. It was just a matter of him getting the minutes before everyone saw what I saw. Luckily I got there first.

Owen

 
At Friday, November 07, 2008 6:57:00 PM, Anonymous Anonymous said...

anymous reggie

he ranks third or fourth nuthing tat special he is good at penetrating thats it he doesnt have a good outside shot i would take nash willams paul over him.

 
At Friday, November 07, 2008 8:44:00 PM, Anonymous Anonymous said...

I'm a Spurs fan, and I don't particularly like Kobe, but I really cannot imagine how Manu's career game could be compared to Kobe's 81.
Statistical analysis misses two very, very important details:

1. Difficulty
Just because Manu went 15-20 doesn't mean he can go 30-40. Receiving the ball and off a good shot attempt is a skill.
Fabricio Oberto went 11-11 once. Why do you think Popovich didn't "go to him till he missed?" Parker's 55 point game was seen as "less amazing"
because it took him 2 OTs? The simple fact that he could sustain his production for 2 OTs is an amazing feat in itself!!

2. Unrecorded stats.
Screens set, good screens set, passes that led to assists, passes that led to open shots, deflections, shots altered, good shot attempts, rebounds that fell into a player's hands,
defensive breakdowns that led to a teammate picking up a foul, loose balls collected, bail-out shots, times a player got in trouble and had to use a timeout,
good plays at opportune times breaking an opponent's momentum and/or forcing timeouts, good cuts that forced the defense to adjust, help-side defense while still guarding his man,
threat to score opening things up for a teammate, how quickly a player gets the opposing player/team in foul trouble/penalty, stat-chasing, end-game intentional fouling,
good box outs that allowed a teammate to get a rebound, hard fouls that send a message to an opponent, number of silly fouls, number of seconds just dribbling the ball,
total time spent just standing around on offense, I could go on and on... and these do not even account for the strength of the other team, playstyle(off. boards vs trans D for instance),
or the schedule(back to back, late season meaningless games, late season playoff pushes, tanking), injuries. Coaching situation also affects a lot of things.
Popovich typically limits TD's minutes and could afford to tinker around with rotation. 90% of the NBA's coaches would probably get fired for that.

Offensive stats do not do a good job of tracking a player's off-ball contributions. That means that a around 75-85% of a player's offensive contributions are not tracked.
Defensive stats are even more lacking. As fascinating as numbers might be, they don't even begin to tell the whole story.

I'm not dissing stats, and I generally agree with a lot of Hollinger's conclusions and he's my favorite statistician.
He also admits to the flaws of his system (last year, his calculations predicted that the Spurs could miss the playoffs).
He said that pre-Allstar Spurs do not equal post-AllStar break Spurs. Most of his numbers make sense when comparing it to observed games.
However, the mere fact that teams do not trade for player X because his PER is higher means that PER can't possibly tell the whole story.
His vaunted point-differential system cannot account for 2 simple little details: 1. Team A's garbage time players vs Team B's. 2. Game significance (must win, cruise mode, etc)
Again, it does give a better indicator of team strength than just raw W-L, but it's not close to absolute.

As for Berri.. Is he the guy who said David Robinson was better than Shaq BECAUSE he had better REGULAR season numbers???
Is he the same guy who said he doesn't need to watch NBA games because stats tell him everything? I could be mistaken though.

Wow.. that was long, sorry to ramble :)

Z

 
At Saturday, November 08, 2008 7:55:00 AM, Blogger David Friedman said...

Owen:

I think that you are reaching a bit in your interpretation of what D'Antoni meant by his remark. You may be right but without directly asking D'Antoni to clarify his statement it could just as easily be supposed that he was making an observation about the limitations of statistics period, not just scoring stats or boxscore stats.

I doubt that D'Antoni made his decision to start Duhon or bench Marbury based on the numbers that you cited. The bottom line with Marbury is that he is a loser. I've advocated for quite some time that the Knicks should simply waive him. D'Antoni coached him briefly in Phx, so I think he saw enough to know he can't win with Marbury running the show. Duhon is a solid pg but--most importantly--he is coachable and he is not going to do anything stupid or selfish.

I wouldn't judge any player solely by numbers but, in general, I think that a 20-10 (ppg-rpg) player is going to be more valuable than a 20-8 (ppg-apg) player. I suspect that the reason for this is that the awarding of assists is more subjective than the awarding of rebounds. Can you name a legit, perennial 20-10 guy who is not a very good player? Randolph has been a 20-10 guy twice in seven seasons but I would argue that he was a fairly productive (though flawed) player during those seasons.

Ranking the greatest single game performance of the past seven years is not a major interest of mine; the reason that I cited Hollinger's work in that regard is to introduce a discussion of his PER rankings of Parker, which I think are skewed and which demonstrate the limitations of that type of analysis. Perhaps WoW has a better handle on Parker's value than Hollinger does but there are also problems with WoW, as I have mentioned before.

The one thing that I do have to say about these single game performances is that it is amazingly difficult to score 81 points in any game, let alone an NBA game. That is why only one other player has done it in 60+ years--and many would argue that that player (Wilt Chamberlain) was the greatest player of all-time. Kobe not only scored 81 points, but he did so with an excellent shooting percentage and he carried his team from a double digit deficit to a victory. I've played my fair share of rec league basketball and relative to that level of play I am in excellent shape and I am an above average shooter; the most points that I have ever scored in a game is 39--albeit in a 40 minute contest, not 48 minutes--and that was with the benefit of making a lot of three pointers. It takes an extraordinary amount of energy to get open that much and to make those shots, particularly if you are exerting any kind of effort whatsoever in other aspects of the game such as defense and rebounding. I don't think that on my best night I could score much more than 40, so to score 81 in an NBA game is simply amazing to me.

As for Landry versus Pietrus, the Magic signed Pietrus to fill a specific need: defense at the shooting guard position. Even if Landry is a better or more productive player, he is not a shooting guard. Perhaps time will prove that you are right that they should have signed him instead but I understand why the Magic signed Pietrus and I think that Pietrus can help them.

As I said, I like Calderon; I just don't rate him among the very top pgs yet and that is why I did not mention him in the earlier post or in this one. He may very well play his way into that discussion by the end of this season.

 
At Saturday, November 08, 2008 8:05:00 AM, Blogger David Friedman said...

Z:

You make some excellent points and cite some very good examples to back them up. I've covered some of that ground in various posts but you summarized the issues very nicely.

I think that Hollinger understands the limitations of his formulas more than some of the fans who slavishly cite his numbers. My point with this post is that it should be pretty clear that as great as Parker played in the first four games he did not perform at a level significantly higher than the career norms of MJ and Wilt. Anyone who swears by PER should stop to wonder what it is that PER is not capturing. Yes, this is a small sample size and by the end of the year PER will not likely be saying that Parker is better than MJ--but why can't PER more accurately assess these four games? If PER comes up with such a screwy value for these games, what else is it measuring imprecisely?

Yes, Berri made the case that D-Rob was better than Shaq and yes he has insisted that he doesn't have to watch NBA games to understand them; in fact, he went further and said that watching games actually biases the observer. As I have said before, I don't buy that premise and I am skeptical of a lot of Berri's conclusions; ironically, sometimes Berri and I make similar evaluations of players but for completely different reasons.

 
At Wednesday, November 12, 2008 2:48:00 AM, Blogger vednam said...

I think you hit the nail on the head when you mentioned how San Antonio's system restricts individual numbers. I think Parker could be a 25-8 guy if he needed to be (just like Duncan could score 27 ppg instead of 18-10).

 
At Wednesday, November 12, 2008 5:23:00 AM, Blogger David Friedman said...

Vednam:

That is why I don't use stats to make my final determination of a player's value. As I noted, I ranked Parker as an elite pg last year and did not have to see him score 55 or lead the league in scoring in the early going this season to understand just how good he is. That is also why my evaluation of Kobe's value has not changed as his scoring average has decreased in recent seasons; his ppg average is being determined by his role on the team not by any decline in his basic skill set.

 

Post a Comment

Links to this post:

Create a Link

<< Home