Kyrie Irving

[wuck]

I hate this constantly off-topic thread. Thinking about closing it until the post-season.

 

[Ohio]

I hate this constantly off-topic thread. Thinking about closing it until the post-season.

Think we could just use a new one.

 

[Simon]

I hate this constantly off-topic thread. Thinking about closing it until the post-season.

Please. You'll be doing everyone a favor.

 

[Lee]

Is someone seriously comparing Korver to Kyrie and saying he is just as good?

Even Gordon Gund can see that is not even close to being true.

 

[Jon]

Then why do you immediately go to that stat and that stat only when comparing individuals?

As opposed to what? We're all aware of the box score stats. +/- tells you what you're not seeing in the box score.

If you think +/- gives a better understanding of the individual than PER than you misunderstand these stats more than I previously thought.

Like I said, we're all well aware of the box score stats (from which PER is derived), but basketball isn't an individual game. I'm not interested in a player who's going to get me '30' if there isn't a way my team can actually win when he's on the floor.

When it's the last 5 minutes of a 30 point blowout, and I'm chilling on the court while my teammate goes on a 10-0 run by himself, giving me a +10, is that really looking at everything?

My teammate scores all the points, I stand around and do nothing, while the whole unit contributes on defense. +/- rates me the same as my teammate that scores the points. That is looking at everything to you? That is an accurate measure of me & my teammate - we are equals?

Yes, that's collecting everything the team did. Your work on defense was noted, and your wise choice to get out of the way of your teammate on offense was noted as well.

But again, remember I keep pointing to observation and lineups. If someone pointed out that the bulk of your plus minutes were during the garbage time of blow outs we'd have a reason to re-evaluate.

(and in trying to discredit my love for +/-, you actually brought up an interesting idea which could be applied to any stat - which is to be able to filter it by "meaningful minutes").

I get it, and there is certainly a place for +/-, particularly in trying to find individuals and units on a given team that play well together.

Where it is terrible is when it's used to compare how good individual players are. Actually, terrible isn't the right word, inapplicable is. On one hand it seems now that you understand this, but on the other hand you continue to compare how good individual players are this way.

So, what gives?

It's an analysis tool, not a ranking tool. And really, that's how all stats should be treated. Getting back to my PER example (and I realize there are a ton of other stats favored over PER these days) it does a decent job ranking players. We know this. It's what it was designed to do.

But drill down to an individual player.

PER can only do it's job correctly if the player we're looking at is roughly as good in areas that PER doesn't track as in the areas that PER does track. It's a reasonable assumption for a lot of players, sure, but what if the specific case we want to consider is an outlier?

How can you tell?

The problem here is you can't and what's even worse is you don't even know you should be looking.

So, Stephon Marbury leads his team to the lottery another season while Jason Kidd leads his team to the playoffs ... and all PER has to tell us about that is ... Huh?!?

Ok, but now we've got some better stats like WS, DWS, BPM and VORP. And why are they better? Well, because they introduce team performance in to the stat. The problem is they credit players based on their box score stats. A great defender might be a great rebounder ... or maybe he's not. The stat is telling us what we want to hear, not what we need to hear.

Or we can go with the latest super charged version of +/-. Is that RPM this week? Hey, cool, they automatically try to account for a player playing with better talent .vs. worse talent. Handy stuff as long as there's enough data to sort that all out... but again, what if it's wrong? How could we tell?

+/- otoh slams you right in the mouth, and if you disagree with what it's telling you, you'd better go look for an answer. Fortunately, we don't have to reverse engineer complex formulas or regressions, we just have to look at the 5-man units that fed the final numbers - and account for who are the players that drove the stats .vs. those who stood around and benefited.

Of course 82games is not just a site that publishes +/-, they also publish a ton of other useful stats going back to when Roland Beech started tracking all this stuff in 2002. I especially like the other on-court and off-court stats they track.

Enough?

 

[shoes22]

You're just not taking criticism well while Gouri seems to think he's participating in a political debate, constantly looking for ways to put words in his "opponent's" mouth so he can jump on them, turn them in to a "meme", and build his reputation with his base.

I hope you realize how ridiculous this sounds...

 

[gourimoko]

 

[Jon]

I hope you realize how ridiculous this sounds...

Yeah, I know. You'd think he'd have better things to do with his time than practice dirty debate tactics on a basketball forum.

 

[gourimoko]

Yeah, I know. You'd think he'd have better things to do with his time than practice dirty debate tactics on a basketball forum.

Jon.. can you explain how you disassociate the player from the team in the +/- lineup/stint data and account for shared time between players playing within groups?

Can you walk me through your methodology; mathematically.

Lastly, can you demonstrate (statistically) how/why Kyle Korver is better than OR has greater impact than Paul Millsap on the Hawks in 2015-16 (Regular Season/Playoffs)?

Thanks.

 

[The Process God]

You're awful, Jon.












And, gouri, I've never seen someone get so baited by a troll as you.

 

[gourimoko]

You're awful, Jon.


And, gouri, I've never seen someone get so baited by a troll as you.

I tried...

I keep telling myself to stop feeding the troll... It'll be one of my New Years Resolutions!

 

[brownsbuck6]

Thread doesn't need closed or locked, it just needs a particular poster banned from it. The last week in here is just unreadable. Ignoring said poster doesn't help, either.

 

[Ohio]

As opposed to what? We're all aware of the box score stats. +/- tells you what you're not seeing in the box score.



Like I said, we're all well aware of the box score stats (from which PER is derived), but basketball isn't an individual game. I'm not interested in a player who's going to get me '30' if there isn't a way my team can actually win when he's on the floor.



Yes, that's collecting everything the team did. Your work on defense was noted, and your wise choice to get out of the way of your teammate on offense was noted as well.

But again, remember I keep pointing to observation and lineups. If someone pointed out that the bulk of your plus minutes were during the garbage time of blow outs we'd have a reason to re-evaluate.

(and in trying to discredit my love for +/-, you actually brought up an interesting idea which could be applied to any stat - which is to be able to filter it by "meaningful minutes").



It's an analysis tool, not a ranking tool. And really, that's how all stats should be treated. Getting back to my PER example (and I realize there are a ton of other stats favored over PER these days) it does a decent job ranking players. We know this. It's what it was designed to do.

But drill down to an individual player.

PER can only do it's job correctly if the player we're looking at is roughly as good in areas that PER doesn't track as in the areas that PER does track. It's a reasonable assumption for a lot of players, sure, but what if the specific case we want to consider is an outlier?

How can you tell?

The problem here is you can't and what's even worse is you don't even know you should be looking.

So, Stephon Marbury leads his team to the lottery another season while Jason Kidd leads his team to the playoffs ... and all PER has to tell us about that is ... Huh?!?

Ok, but now we've got some better stats like WS, DWS, BPM and VORP. And why are they better? Well, because they introduce team performance in to the stat. The problem is they credit players based on their box score stats. A great defender might be a great rebounder ... or maybe he's not. The stat is telling us what we want to hear, not what we need to hear.

Or we can go with the latest super charged version of +/-. Is that RPM this week? Hey, cool, they automatically try to account for a player playing with better talent .vs. worse talent. Handy stuff as long as there's enough data to sort that all out... but again, what if it's wrong? How could we tell?

+/- otoh slams you right in the mouth, and if you disagree with what it's telling you, you'd better go look for an answer. Fortunately, we don't have to reverse engineer complex formulas or regressions, we just have to look at the 5-man units that fed the final numbers - and account for who are the players that drove the stats .vs. those who stood around and benefited.

Of course 82games is not just a site that publishes +/-, they also publish a ton of other useful stats going back to when Roland Beech started tracking all this stuff in 2002. I especially like the other on-court and off-court stats they track.

Enough?

Your point of looking at the 5 man lineups is a good one - and at least that is taking data from 10 players and applying it to 5 instead of just 1.

But...what about the 5 players on the other team?

Do you not see how easily this stat can be skewed?

 

[Jon]

Your point of looking at the 5 man lineups is a good one - and at least that is taking data from 10 players and applying it to 5 instead of just 1.

But...what about the 5 players on the other team?

Do you not see how easily this stat can be skewed?

That's true, but it's just another reason why knowing your team is important.

Popcornmachine.net is another of my favorite sites because of how they track the game flows and rotations.

When you follow the rotations closely on a game to game basis, you constantly test all those concerns you're expressing.

And sometimes the season gives us great snapshots to examine a player.

For instance, when Delly was forced to start at the beginning of the season, we got a long look at how he'd do against NBA starters and not just reserve units.

So, like I've said from the beginning, I could be missing something about Korver, but there's a LOT of data there to say he's really important to the Hawks. And not just his shooting, but his presence on the floor within their system.

But you'd have to say the same thing when comparing box-score states. They look great? But they could be missing something.

The thing is, more data makes Plus Minus stronger. It helps to eliminate those exceptions that could skew the result. But another season of Stephon Marbury averaging 20 & 8 isn't going to tell you those numbers aren't contributing to winning.

Stats that cause you to question your presumptions are a wonderful thing. The act of dismissing them (or failing to dismiss them) is instructive.

Just because none of us want to dig in deep enough to do just that, doesn't make it wrong. But what if it's right?

Well, Korver is getting older and it looks like he's already fallen off some in the past year. If opponents stop struggling to defend him or he gets hurt, we should see the Hawks make a precipitous fall that the addition of Howard will do nothing to prevent.

But I suppose fans will just blame that on Howard...

 

[David.]

I hate to even ask but why are people saying Rubio is "fucking garbage"?

Just because he lacks svoring efficiency? Because he may be the top pg defender, top 5 in assist ratio and third in ast/to, and a good rebounder for his position.

Can we start to fix some of these biases and misconceptions?