Smart 9th in real +/- among PGs

[Celtics18]

Yes, I've seen this. It is a non-technical summary, not a real methodology description. The actual methodology is not published.

Honestly, Koz, I ask this not as a wise-ass or as a contrarian, but I'd really like to know which stats are trustworthy?

Which ones tell us, objectively and beyond criticism, who are the best players in the game?

[kozlodoev]

For what it is worth I think this is better than PER. But my opinion is worth somewhere between a turd and a bag of chips also.

PER is a great statistic if you know what it represents. It's pretty much a per-minute weighted average of a player's box score statistics -- so it's a quick answer to the question which player is more productive on the court. It's downfall is that (a) values extrapolated from low mpg may be severely biased, and (b) people tend to use it as a be-all, end-all measure of player ability. It isn't that.

RPM, on the other hand, opens the door for all sorts of estimation biases in the model estimation, so the question about its worth goes beyond how it is interpreted.

[kozlodoev]

Yes, I've seen this. It is a non-technical summary, not a real methodology description. The actual methodology is not published.

Honestly, Koz, I ask this not as a wise-ass or as a contrarian, but I'd really like to know which stats are trustworthy?

Which ones tell us, objectively and beyond criticism, who are the best players in the game?

My point of view is that to know how much to trust a stat, one has to understand how it is constructed in order to be able to assess its limitations.

Most people instinctively have this type of understanding for simple stats such as raw totals, per-game averages and per-minute/per-possession averages. It is not so simple with relatively nontrivial statistical estimations such as RPM.

[Celtics18]

Yes, I've seen this. It is a non-technical summary, not a real methodology description. The actual methodology is not published.

Honestly, Koz, I ask this not as a wise-ass or as a contrarian, but I'd really like to know which stats are trustworthy?

Which ones tell us, objectively and beyond criticism, who are the best players in the game?

My point of view is that to know how much to trust a stat, one has to understand how it is constructed in order to be able to assess its limitations.

Most people instinctively have this type of understanding for simple stats such as raw totals, per-game averages and per-minute/per-possession averages. It is not so simple with relatively nontrivial statistical estimations such as RPM.

Fair enough.  I'm not really a big fan of stats that I don't understand either. 

[byennie]

It's worth keeping note that the OP posted 6 different (albeit related) metrics, not just RPM.

[Celtics18]

Regardless of the validity of the stats posted, I will say I have been extremely impressed with Marcus Smart's rookie season. 

Let's remember that Marcus just turned twenty-one about a month ago.  I think we tend to overlook how young he is a). because he went back to school for his sophomore year, and b).  because he already plays like such a veteran.

The fact that he's light years ahead of a vast majority of other players his age already in the NBA on the defensive end seems to work against him as far as perception for his future goes.  When we watch Giannis Antetokounmpo, Andrew Wiggins, and Nerlens Noel play, we have a tendency to get excited for their futures, not because of the players they are now, but because of the players we project them to become.

When we look at Marcus Smart, we see someone playing like a ten year veteran already.  This makes us think this is how he will be playing in ten years.  It's harder to allow for the fact that as a twenty-one year old kid, he might very well make some significant improvements to his game as well.

No, he's not a super athlete.  But, he is strong as an ox, with very strong, quick hands, and excellent lateral movement. 

Perhaps what has impressed me the most about him, though, is his shooting stroke from behind the arc.  Obviously he's not an elite shooter, but coming in to the draft, his outside touch was widely reputed to be one of his significant drawbacks.  It hasn't been.  He's been very reliable and serviceable from three point range. 

The fact that his shooting has improved as much as it has as early as it has makes me believe that he will continue to get better as a shooter. 

I do think that Marcus Smart has the chance to become a star player in the NBA.  What's even better is that he'll become the kind of star that will fit the gritty, blue collar culture of our franchise.  He'll never be a flashy, high flying scorer, who overpowers his opposition with speed and above the rim athleticism.  I think that's a given. 

What he may well be is a very heady guard who plays tenaciously on both ends of the floor and has a knack for making big plays.

[oldtype]

For what it is worth I think this is better than PER. But my opinion is worth somewhere between a turd and a bag of chips also.

PER is a great statistic if you know what it represents. It's pretty much a per-minute weighted average of a player's box score statistics -- so it's a quick answer to the question which player is more productive on the court. It's downfall is that (a) values extrapolated from low mpg may be severely biased, and (b) people tend to use it as a be-all, end-all measure of player ability. It isn't that.

RPM, on the other hand, opens the door for all sorts of estimation biases in the model estimation, so the question about its worth goes beyond how it is interpreted.

RPM is just a player's on-court +- stat controlled for the +- of his teammates. It's based on real numbers just as much as PER is.

Both stats have their flaws due to inherent biases (like the ones you described for PER), but I don't see how you can argue that one is more "open to interpretation" than the other. (If anything, PER is a much more "fuzzy" stat because the weights for each individual stat is being arbitrarily assigned.)

EDIT: never mind actually, I see somebody already brought this up on the last page

[Boris Badenov]

"Celtics replace statistically puzzling veteran PG with statistically puzzling rookie PG."

[PhoSita]

"Celtics replace statistically puzzling veteran PG with statistically puzzling rookie PG."

There's nothing all that puzzling about Smart's stats.

He's the best defender in this rookie class, and already one of the best at the point guard position.

On offense, he's been allowed to operate mostly off the ball, and because he's been able to hit the three at close to a break-even rate he's not a detriment on that end the way most offensively inept rookies are.


If anything, the most statistically puzzling player on the Celts is Olynyk, who is fairly polarizing due to his appearance and style of play, and confounds most attempts to explain through statistics or intangible explanations alone what his true on-court value is.  Sound familiar?

[Boris Badenov]

"Celtics replace statistically puzzling veteran PG with statistically puzzling rookie PG."

There's nothing all that puzzling about Smart's stats.

He's the best defender in this rookie class, and already one of the best at the point guard position.

On offense, he's been allowed to operate mostly off the ball, and because he's been able to hit the three at close to a break-even rate he's not a detriment on that end the way most offensively inept rookies are.

I didn't mean you can't learn what kind of player he is from his stats. I meant that his profile as a player is like Rondo's, in that (a) he's far from a stereotypical PG, and (b) he's exceptionally good at some things and really bad at others, at least at this point in his career.

I do agree with you that KO is...similarly different.

[CelticGuardian]

Smart as he stands now, is a hustle player/defensive stopper. I think we can only hope he develops into a Kyle Lowry kind of player. I think the Westbrook comparisons were extremely overblown... he's explosive, no doubt. But come on, guys...

[Boris Badenov]

"Celtics replace statistically puzzling veteran PG with statistically puzzling rookie PG."

On offense, he's been allowed to operate mostly off the ball, and because he's been able to hit the three at close to a break-even rate he's not a detriment on that end the way most offensively inept rookies are.

Just following on your point about threes, it's a great example of how Stevens adapted a player's role to minimize his weaknesses. You could imagine Smart superficially putting up better numbers if he handled the ball and/or drove more, but he'd probably be much less efficient.

[greece66]

Smart as he stands now, is a hustle player/defensive stopper. I think we can only hope he develops into a Kyle Lowry kind of player. I think the Westbrook comparisons were extremely overblown... he's explosive, no doubt. But come on, guys...

What Westbrook comparisons? (honest question)