OT: Stats indicate Sergio actually good

...says Wayne Winston:

Whaddya know?

 

The people who don't like what he says?

 

People spend about two weeks of their lives at traffic lights

 

if the majority of sergio's minutes came during a time that was meaningless to the outcome of the game, how the team performed in those minutes is basically meaningless.

 

I feel like there already may be a thread about this.

On the plus side, drexlersdad just made this experience much better.

 

Can't you read?

The Kings play 4 points worse than average for the season. The sad thing is they traded Rodriguez, who had played great. When Sergio was in the Kings were an amazing 9 points better than average(after adjusting for strength of opponents). When Sergio was out Kings played 7 points worse than average.

 

Pretty small sample to be using +/- meaningfully. Especially since, as rocketeer pointed out, the majority of his minutes were played in garbage time.

What you bolded are facts. What is not fact is the predictive value of those facts and whether the indicate Rodriguez's "true talent" level.

 

Can the mods please make an official Sergio thread and just merge every thread about him with that thread...I don't care about him enough to keep seeing threads about him, and I'm sure I'm not alone.

 

If his stat is the best, he should be a one-trick pony. Anything else would be just using inferior data.

Why would you use other stats unless PER doesn't really measure what's important? (Which, of course, it doesn't.)

If you said are better, then I would agree with you. But have been better? Where is it that they've supposedly been better? In that small sample, right? Are you saying that if Sergio were to spontaneously explode, it would be impossible to say how well he played in that small sample of games? That we can only know how he played in those games after he's played in other games?

Now, presumably you have in mind that with a stat like this he might be getting the credit for other players that he routinely plays with. But that's not a small sample size problem (because if he played in a million games with exactly the same people, you'd have the same problem) that's a non-varied sample problem. But, unless you have a bunch of players all with identical scores, there has already been variety.

And on what basis do you think that they are "crazy"? Presumably because you disagree with them. But that could just be a sign of you overlooking important features. James Jones, for example, seemed to help the Blazers win a lot more than he should have, given his limited skills. There are two possible responses: "he can't be helping, so the stat is wrong!" or "maybe there are subtle ways that a player can help that I'm not seeing". Unless you're Tex Winter or something, I think the latter is quite likely. And in Jones' case, it probably had a lot to do with being calm, being more of a veteran than his teammates, and in talking a lot on D - all of which his teammates said was true of him. Does that make him a great player? No. Does that mean he'll be useful to all teams in the same way? Of course not. Does that mean he was good for the Blazers? Absolutely, and in a way that PER can't measure.

 

True. But how do you know that it did? The weird thing about Sergio in Sacramento is that he seemed either to be starting (particularly when they had injuries) or not playing at all. (Which might be why at least three Kings players, including the sainted Tyreke Evans, have been publicly bitching about Westphal's rotations lately.) Prove to me that the bulk of Rodriguez's minutes were "meaningless".

 

Are you giving up on the holy grail of unified field theory?

I see Winston's job as to make his statistic as fine-tuned as possible. Taken to extremes, your view would advocate against any "combined" stat - all we need is the raw data and a commentator to talk about all of it. But essentially that's what a stat-hound tries to do: combine the data in revealing ways. (That's why it's adjusted +/-.)

To push your analogy, think of Winston as a banana-supplier. You don't want him messing around with apples, you want him producing the perfect banana. You don't have to eat it, but you want it to exist. That's when it's good to be a "one-trick pony".

 

The problem is, he assumes more points is necessarily better. But he doesn't look at the whole picture on stats. How many were being scored against them? What was their win-loss record during that period? Until you look at the big picture instead of just stats, you have an imperfect model to base your opinion off of.

 

Actually the numbers do indicate that Sergio was good for the Kings from all these aspects - the scored more when he was on the court than when he was not, they gave up less points when he was on the court and they won 55% of the time he was on the court (which is very nice for a team with a 31% win rate).

It is true, however, that the numbers are for only 39 games and Sergio averaged less minutes per game in Sacramento than he did in Portland - so, I would not be quick to point them as "definitive" - but it is hard to argue with the conclusion that he was good for Sacramento when they did play him.

Is it because Sacramento was smart enough to only play him when he would not be a liability, or is it because he was really good there - I have no idea, I did not watch him enough there to tell you for sure - but, Sergio definitely had his best stretch of NBA basketball in the small amount of time he played for Sacramento.

At the end of the day - he did not fit in Portland at all, his style did not mesh with Roy at all to the point that Miller fits with Roy much better than he did - and hopefully he gets all the rope in the world in NYK to swim or sink - I got nothing against Sergio - and wish him all the best. He seems like a very nice guy. Not the world beater we were told he is, but in the right system, I think he can succeed. Luckily for him - he gets a chance to prove it in the best possible scenario for him.

 

Wayne Winston is a man who believes statistical models are truth. He is more certain about the numbers his system produces then I am about anything.

I am extremely leery of folks who believe their own hype that much. Especially statisticians who believe their own models represent facts instead of just being really good guesses.

 

It might be helpful to think of statistics as summaries. They help you see "a" big picture, but they leave out details. The average human would be a weird mixture of males and females. Such a creature rarely exists. That would be one extreme misuse of a summary statistic like average. On the other hand, if someone produces 20 points a game nearly every game, it is quite fair to expect them to perform similarly in the future.

As many have said in many threads here, if the sample size is large enough, the exceptions lose importance and the summary is more predictive.

And most of you knew this already, but it seems like we argue about statistics and sample sizes a lot in here.

 

sergio didn't start a single game for the kings, so i'm not sure exactly where you got the idea that he either started or didn't play at all.

and sergio only played 8% of the kings' minutes that game during the 4th quarter or overtime with 5 minutes or less remaining while the game was within 5 points. that doesn't prove that the majority of sergio's minutes game during garbage time, but it does prove that the kings didn't want him on the court during the minutes that mattered most.