The Chicago Bulls Analytics Thread

[Dr Spaceman]

Yes, it is.

Spaceman, you should start up a Chicago Bulls statistics thread and moderate the proceedings. You not only know your stuff, but you know how to communicate it too.

Generally I'm someone who hangs out on the PC Board, but I saw a thread on here asking about ESPN/Engelmann's Real Plus Minus and thought I'd clear some things up. Got a few requests to start and maintain/moderate an analytics thread, and I like the idea so I'm on board.

I'll use this thread to share statistics that I've come across and find useful, and to answer questions about the development, interpretation, and correct usage of statistics. Hopefully this thread can also serve as a discussion point for analytics as a whole, as I think there's something to be gained from the movement even amongst those who aren't statistically inclined.

Remember, analytics is about more than stats. It is about contextualizing, quantifying, analyzing, and most importantly understanding the game of basketball. Statistics are a big part of that, but so is watching film and deepening your understanding of the areas of the game that have nothing to do with numbers.

Everyone is welcome to participate, even if your knowledge of analytics is zero. Ask questions, share new things you've found, even develop your own stats if you're so inclined. Just keep an open mind is all I ask.

Where to start:

RAPM (Ridge Regressed Adjusted Plus/Minus)

Layman's explanation:

Spoiler:

So I'll say this:

Stats in the (+/-) family are not meant to be taken as "player x is better than player y". At the heart, these statistics are meant to measure "when player x steps on the court, his team improves by x amount". This is an absolutely crucial distinction to make, because so often people look at stats like this and say "No way are Manu Ginobili and Amir Johnson better than Kevin Durant, this is B.S." So this isn't a stat you can just look at and stop thinking, in fact these stats are designed to inform your thinking. People who get really into plus minus stats (like myself) do so because the numbers that are spit out match pretty much exactly what we see on film.

So what is this measuring? Basically, the idea is that every player in the NBA plays in a multitude of lineups every season and that using all of this lineup data, you can measure the precise impact of each individual player on the scoring margin.. Think of it like an algebra problem, where

scoring margin= Bulls1+Bulls2+Bulls3+Bulls4+Bulls5-Jazz1-Jazz2-Jazz3-Jazz4-Jazz5

If you do this over thousands of lineups and thousands of opponent lineups, you can start to get a good picture of each of the individual variables in this equation. (Note here that Bulls 1,2,3 would be Jordan, Pippen, Rodman, etc.)

The important thing to remember is that you should take these number for exactly what they are: a given player's impact on the scoring margin. Nothing more, nothing less.So for example, I still think Durant is a better player than Ginobili, because his skill set is superior, he plays the #1 role on offense, and he plays far more minutes than Ginobili. I sort of "filter" RAPM data through my own observations to come up with a clear picture of a given player. So the data are telling me that Ginobili, in his role, is one of the most impactful players in the league. But when you consider how limited his role actually is, I'm not likely to rank him very highly overall.

Regarding your questions about specific Bulls players, all I'm going to say is that years and years of data agree that players who play the way Rose has this season do not help their teams win. And players like Mirotic who shoot well and space the floor are far more impactful than you'd think (look at nba.com's lineup data, Mirotic is a part of basically all of the Bulls' best offensive lineups).

2015 Prior Informed RAPM: http://stats-for-the-nba.appspot.com/ratings/RAPM_2015.html

2008-2014 Prior Informed RAPM: http://www.gotbuckets.com/statistics/rapm/2014-rapm/

Doctor MJ's spreadsheet with Prior Informed data from 1998-2012: https://docs.google.com/spreadsheets/d/1h20JYcZJu2tGNIyOwVbNfez0-zXXy5ItLyXC4qTE5D8/edit#gid=0

A more technical explanation of RAPM: http://www.gotbuckets.com/what-is-apm/

Player Tracking, Play Type, & Lineup Data

http://stats.nba.com

These are "scouting report" type stats, found on NBA.com, and their utility is awesome.

Player tracking is awesomely useful for several niche type stats that would not otherwise be known. Rim protection, catch and shoot percentage, etc.

Play type is the old synergy. Most useful information is how many points per posession a player scores in isolation, post up, PNR handler, etc. Helps quantify those observations.

Lineup data is very useful for determining player fit and portability. Stray observations I've found on the Bulls this season: !. Gasol/Noah lineups generally do not play well 2. Dunleavy/Mirotic lineups are SCORCHING teams offensively 3. Mirotic and Brooks show up most often in Bulls best offensive lineups.

Player Tracking Plus Minus

A new development in 2015, this is the first statistical plus minus attempt to use player tracking data in place of the box score. Basically RAPM is "built on" by the player tracking data mentioned above.

https://docs.google.com/spreadsheets/d/1GtCDQw94kpcOw_kPhyH8F5cIjPT3QTsOGqvrX_hMCo8/edit?pli=1#gid=0

http://counting-the-baskets.typepad.com/my-blog/2014/09/introducing-player-tracking-plus-minus.html

Why is this a big deal? As illustrated here: http://apbr.org/metrics/viewtopic.php?f=2&t=8633&start=210#p23003 it is currently outperforming every single other type of metric by a gigantic margin. Nothing else comes close to being able to predict wins. If this continues, we could have by far the best all in one metric ever created in our hands.

We could eventually have a stat here that is able to capture the value of players that typically go unnoticed. Doctor MJ's take:

Spoiler:

To me what the goal of those folks should be is this:

Until we get a PT SPM that can roughly peg any star with a consistent track record's RAPM, we don't have it good enough.

once we truly have it in that ballpark we'll have a truly reliable metric with nearly unbiased validity for the first time, which in and of itself would be huge, but possibly even bigger would be the fact that you'd truly be able to analyze how a guy makes his impact which would make portability estimations much better and thus help scouting and team building in addition to team self diagnostics

That might seem pie in the sky but consider that nowadays we actually have PT based gravity metrics. This stat right now doesn't see Korvers impact, there is data this can eventually use that can grasp even that

Thanks SideshowBob for introducing us to this: http://forums.realgm.com/boards/viewtopic.php?t=1374892.

WOWY

"With or Without You"

A method created by this forum's very own ElGee. Essentially this is a way to estimate players' impact by SRS (point differential) performance in games they've missed. He explains it better than I: http://forums.realgm.com/boards/viewtopic.php?t=1333570

Another way to estimate "impact", and this method also works with ORTG/DRTG on/off (can be found on basketball reference, I can talk more about that too if people are interested.

Final Notes

Here is a dropbox folder where I keep a bunch of spreadsheets you might find useful (and this will become more and more full over time): https://www.dropbox.com/sh/l9csmf8as5yjv1h/AADdREYub8D13pZUDqCblpgWa?dl=0

-When using impact stats, to reduce noise it is helpful to look at very large samples (ie. if one player is slightly above another in 2014 it might not be that big a deal, but if he is consistently better every year from 2007-2014 that's a very big deal)
-Look for impact stats to "agree" (ie. RAPM says a guy is huge impact, team posts terrible record without said player, team falls apart when player goes to the bench) If you have multople sources saying the same thing, believe it.
-don't use any single stat as an "all-in-one" that's not what they're designed for, that's not how it should be used
-don't discount an entire statistic because you don't like what it says

Anyway, this is all I got for now, but hopefully this can stir some useful discussion. Always like to see new minds come around to the analytics movement.

[Dr Spaceman]

My take on Kyle Korver:

Spoiler:

I meant do you believe that Korver, e.g., really has had much more positive impact on the court for Atlanta than Butler has had for the Bulls?

There's a reason I have a picture of Korver as my avatar

You may not realize it, but this is one of the most important questions you could've asked in your quest to understand impact statistics. This is generally the part of the conversation where people tend to look at me sideways or even outright dismiss me, so I really hope you'll read this with an open mind.

Think about this: on offense, a basketball player is always in one of two conditions. Either he has the ball, or he doesn't. A player can contribute by doing good things with the ball (this is what's captured by the box score and what most people focus on when watching games) or he can do good things without the ball (this is something that isn't traditionally measured by any of our counting statistics and something most people don''t notice).

Now with that in mind, consider Korver's shooting. Obviously, people see that he scores 12 ppg and question how this guy can be seriously considered an impact player. But consider a couple things: among players who have equalled Korver's 3 point shooting volume in history, the highest percentage ever recorded was 45%. Korver is currently shooting 49.6%. He's bombing 3s at a rate equivalent to some of the highest in history, and he's hitting them at a percentage no one has ever come close to. On Catch & Shoot plays, per NBA.com, Korver is averaging over 1.3 points per possession. The best offense in NBA history scored about 1.15 points per possession.

Now another thing Korver has mastered is being a brutally quick decision maker. He catches and either shoots if he's open or finds an open teammate to pass to. This means that failed attempts to get him the ball cost very little shot clock time, and he's not taking bad shots when he does possess the ball. This is the most underrated aspect of what made guys like Bird and Jordan so ridiculous on offense. It was always catch, read, attack or pass within the blink of an eye. Korea does this as well as anyone, and the result is he doesn't waste possessions. His teammates who are better at isolating and pick and roll have more shot clock to work with.

And another thing in Korver's favor: right now he leads the league in True Shooting Percentage. What that means in practice is that a shot attempt by Korver on average results in more points than a shot attempt by any other player in the NBA. This is a big, big deal.

Okay so we've established what Korver does with the ball that makes him so special, but what about without the ball? Well, given that he's the best in the league when he actually does decide to shoot, any reasonable opposing coach is going to do their best to keep the ball out of his hands. And this is the secret to the Hawks offensive success. It's almost a joke how far opposing teams will go to stop Korver from shooting. I'm talking doubles at the 3 point line, switching every screen, I've even seen multiple times this year opposing defenses concede an open dunk just to avoid Korver being open. It's ridiculous.

And Korver does not behave like a normal shooter. Watch him during a Hawks game sometime, he's constantly moving, sprinting in circles around the arc, making decisive dives to the corner. He's the biggest threat on the Hawks, and all eyes are on him all the time. Defenses literally treat him like he's LeBron.

We imagine the typical low volume 3 point shooter as a guy who sits in the corner and shoots when he's passed to. But this is not the case with Kyle. The Hawks will run him around screen after screen, watch the defense shatter, and then attack the craters he leaves behind. They are so brutally effective at using Korver's gravity to set up 2-on-1s and mismatch isolations. Korea doesn't get credited with an assist on these plays, but he deserves it.

And think about what you'd do if you're defending a pick-and-roll with Korver lurking on the weak side. You absolutely cannot help off of him. In Thibodeau's defensive scheme, the weak side corner help man needs to drop into the paint and bump the roll man, impeding his way to the basket. But if that defender is guarding Korver, you can't use that guy to help on the pick and roll at all. Watch the Hawks run pick and roll, and it honestly looks like you could drive a semi through the lane at times.

So yeah, Korver is having tangible, transcendent impact every single possession, and he never even has to touch the ball to do so.

So big picture, what does this mean? I don't think Korver is a better basketball player than Jimmy Butler. Butler has much better defensive told, a more versatile skill set, and can be a high-usage guy in a way that Korver can't. However, Jimmy Butler is not having as much impact on the basketball court as Korver is right now. Very few players are.

And this gets a little philosophical, but follow me here: we tend to think of basketball players in terms of how well they could do without quality talent around them. But to me this is a fallacy, because if the goal is to win a championship, I want players that play really well when they have other talented players around them, even if they need other talent to do their thing effectively.

A real world example: The Knicks would be far, far worse with Korver in place of Carmelo. This is a given, since Carmelo is much better suited to carrying a team like that. But flip things around: do you honestly believe the Hawks would be better with Carmelo in place of Korver? Because I legitimately don't. I think the Hawks would be worse with Carmelo.

Another real-world example: You have the option of putting any shooting guard in the league on the Cavaliers. do you pick anyone other than Korver? The guy is going to make life so much easier for the other offensive stars, and when he does use possessions, he's going to use them better than even the other stars would!

Now what you'll take issue with is Korver's context-dependency. And that's totally fair, and for a lot of teams they wouldn't hesitate to take a litany of other guys over Korver. But think about this: is it more valuable to make a bad team mediocre, or to be able to make a great team championship-worthy? Because the latter is what Korver does. And I'll take that over the Carmelo's of the world any day.

Wow that was long. Anyway this is why I'm such a huge advocate of film study. Once you accept RAPM as being a valid indicator of "impact" you start looking at basketball in a different way, and start to understand the unseen impact of guys like Korver. Other big ones are Steve Nash, Bill Walton, David Robinson, etc.

[coldfish]

Not so long ago, players were evaluated heavily just by raw stats. Points, rebounds and assists. The end result was that guys who were relentless chuckers or awful defenders had much better reputations than their contribution justified. When analytics came around, I thought they were awesome. They almost invariably lined up with my "eye test" and they were a great tool to argue with the old school type people still using points, rebounds and assists.

Analytics moved fast though. The old school people became converts rather quickly. The other day Aaron Brooks had 22 points on 23 shots while playing awful defense. A decade ago people would have thought he was the star of the game but that is basically gone. Its to the point where analytics are just confirming what most people see and they almost have little value in a conversation.

Where basketball is interesting to me is the "why" of a player having a +2.3RAPM or a 17.2 PER. The x's and o's of a guy moving around the court and impacting the game. Analytics really just don't say much in that regard.

[DanTown8587]

So big picture, what does this mean? I don't think Korver is a better basketball player than Jimmy Butler. Butler has much better defensive told, a more versatile skill set, and can be a high-usage guy in a way that Korver can't. However, Jimmy Butler is not having as much impact on the basketball court as Korver is right now. Very few players are.

This is where basketball and baseball analytics (the father of sports adopting analytics) have one major difference. In baseball we can say one hitter is better than the other because of certain stats such as OBP, OPS, etc. In pitching we can say one pitcher is better than the other because of a lower FIP or some other stat. In basketball, because it's five man units, it's impossible (literally) to quantify one player and then openly compare him to another player with different skill sets being asked to do different things.

In basketball, the best teams maximize the system they have at that time to fit their talents. The Hawks (and the Spurs before them) are examples of taking players that have some minor flaw but creating a system where that flaw isn't able to be extracted. So when someone tells me that Korver has a higher impact than Butler, I realize they've fallen into a fallacy of believing that stats CAN tell the whole story and in the NBA that simply isn't true. Sure Korver does things with shooting that are high valuable and create value to his team; however, that doesn't account for the fact that Butler has a different skill set. So while Atlanta wouldn't run Jimmy off screens, they could if they wanted have two viable ball handlers and create a system that plays inside out with cutters and post players taking advantage and running multiple P*R actions, similar to what San Antonio did in the mid 2000s when they were winning. And we haven't even gotten into the different things Atlanta could do defensively with Butler or the different ways they can play in transition.

And by the way I'm not saying that Butler IS more impactful than Korver or vice versa but rather the idea that you can quantify one's impact and compare it against another player is practically impossible. Both have impacts that are drastically different; without literally placing Butler on the Hawks and Korver on the Bulls for a sizable sample, we'll simply never know. Because at the end of the day, analytics tell me how much better Korver makes the Hawks and how much better Jimmy makes the Bulls but that doesn't mean you can pull them out and say "well higher RPM/VORP means better player". I think some (not saying you) are hoping that one day basketball gets to the point of baseball where basically a guy with a spreadsheet could tell you the best players in the league but that's simply impossible in the NBA.

[Dr Spaceman]

So big picture, what does this mean? I don't think Korver is a better basketball player than Jimmy Butler. Butler has much better defensive told, a more versatile skill set, and can be a high-usage guy in a way that Korver can't. However, Jimmy Butler is not having as much impact on the basketball court as Korver is right now. Very few players are.

This is where basketball and baseball analytics (the father of sports adopting analytics) have one major difference. In baseball we can say one hitter is better than the other because of certain stats such as OBP, OPS, etc. In pitching we can say one pitcher is better than the other because of a lower FIP or some other stat. In basketball, because it's five man units, it's impossible (literally) to quantify one player and then openly compare him to another player with different skill sets being asked to do different things.

In basketball, the best teams maximize the system they have at that time to fit their talents. The Hawks (and the Spurs before them) are examples of taking players that have some minor flaw but creating a system where that flaw isn't able to be extracted. So when someone tells me that Korver has a higher impact than Butler, I realize they've fallen into a fallacy of believing that stats CAN tell the whole story and in the NBA that simply isn't true. Sure Korver does things with shooting that are high valuable and create value to his team; however, that doesn't account for the fact that Butler has a different skill set. So while Atlanta wouldn't run Jimmy off screens, they could if they wanted have two viable ball handlers and create a system that plays inside out with cutters and post players taking advantage and running multiple P*R actions, similar to what San Antonio did in the mid 2000s when they were winning. And we haven't even gotten into the different things Atlanta could do defensively with Butler or the different ways they can play in transition.

And by the way I'm not saying that Butler IS more impactful than Korver or vice versa but rather the idea that you can quantify one's impact and compare it against another player is practically impossible. Both have impacts that are drastically different; without literally placing Butler on the Hawks and Korver on the Bulls for a sizable sample, we'll simply never know. Because at the end of the day, analytics tell me how much better Korver makes the Hawks and how much better Jimmy makes the Bulls but that doesn't mean you can pull them out and say "well higher RPM/VORP means better player". I think some (not saying you) are hoping that one day basketball gets to the point of baseball where basically a guy with a spreadsheet could tell you the best players in the league but that's simply impossible in the NBA.

Good post . Thanks for sharing.

Part of why I wanted to do this is because I explicitly don't believe that basketball is quantifiable like baseball. My desire is to teach people the uses, but also just as importantly the limitation of these stats.

My taking Korver over Butler is a judgement call. That's the beauty of basketball and why it's different.

Statistics are evidence. Watching games and film is also evidence. What you're doing in talking about skillsets and strategies is even more evidence. What I want to do is teach people how to use and balance this evidence in their judgements. That's analytics, it's not just stats.

[Dr Spaceman]

Not so long ago, players were evaluated heavily just by raw stats. Points, rebounds and assists. The end result was that guys who were relentless chuckers or awful defenders had much better reputations than their contribution justified. When analytics came around, I thought they were awesome. They almost invariably lined up with my "eye test" and they were a great tool to argue with the old school type people still using points, rebounds and assists.

Analytics moved fast though. The old school people became converts rather quickly. The other day Aaron Brooks had 22 points on 23 shots while playing awful defense. A decade ago people would have thought he was the star of the game but that is basically gone. Its to the point where analytics are just confirming what most people see and they almost have little value in a conversation.

Where basketball is interesting to me is the "why" of a player having a +2.3RAPM or a 17.2 PER. The x's and o's of a guy moving around the court and impacting the game. Analytics really just don't say much in that regard.

I'm definitively not pushing stats as a tell all here. As advanced as the movement has become, most people still have relatively little understanding of these stats and what they mean and represent. (See recent thread asking WTF is real plus minus). Plus we're basically just at the tip of the iceberg with how deep this stuff can go, player tracking and plus minus data is still really new.

Also, stats!= Analytics. It's more than that.

[DanTown8587]

So big picture, what does this mean? I don't think Korver is a better basketball player than Jimmy Butler. Butler has much better defensive told, a more versatile skill set, and can be a high-usage guy in a way that Korver can't. However, Jimmy Butler is not having as much impact on the basketball court as Korver is right now. Very few players are.

This is where basketball and baseball analytics (the father of sports adopting analytics) have one major difference. In baseball we can say one hitter is better than the other because of certain stats such as OBP, OPS, etc. In pitching we can say one pitcher is better than the other because of a lower FIP or some other stat. In basketball, because it's five man units, it's impossible (literally) to quantify one player and then openly compare him to another player with different skill sets being asked to do different things.

In basketball, the best teams maximize the system they have at that time to fit their talents. The Hawks (and the Spurs before them) are examples of taking players that have some minor flaw but creating a system where that flaw isn't able to be extracted. So when someone tells me that Korver has a higher impact than Butler, I realize they've fallen into a fallacy of believing that stats CAN tell the whole story and in the NBA that simply isn't true. Sure Korver does things with shooting that are high valuable and create value to his team; however, that doesn't account for the fact that Butler has a different skill set. So while Atlanta wouldn't run Jimmy off screens, they could if they wanted have two viable ball handlers and create a system that plays inside out with cutters and post players taking advantage and running multiple P*R actions, similar to what San Antonio did in the mid 2000s when they were winning. And we haven't even gotten into the different things Atlanta could do defensively with Butler or the different ways they can play in transition.

And by the way I'm not saying that Butler IS more impactful than Korver or vice versa but rather the idea that you can quantify one's impact and compare it against another player is practically impossible. Both have impacts that are drastically different; without literally placing Butler on the Hawks and Korver on the Bulls for a sizable sample, we'll simply never know. Because at the end of the day, analytics tell me how much better Korver makes the Hawks and how much better Jimmy makes the Bulls but that doesn't mean you can pull them out and say "well higher RPM/VORP means better player". I think some (not saying you) are hoping that one day basketball gets to the point of baseball where basically a guy with a spreadsheet could tell you the best players in the league but that's simply impossible in the NBA.

Good post . Thanks for sharing.

Part of why I wanted to do this is because I explicitly don't believe that basketball is quantifiable like baseball. My desire is to teach people the uses, but also just as importantly the limitation of these stats.

My taking Korver over Butler is a judgement call. That's the beauty of basketball and why it's different.

Statistics are evidence. Watching games and film is also evidence. What you're doing in talking about skillsets and strategies is even more evidence. What I want to do is teach people how to use and balance this evidence in their judgements. That's analytics, it's not just stats.

Hey, if that sounded attacking at all that wasn't my intention, I just wanted to focus on one part of what you said and add how I feel analytics are important to today's basketball and to how we talk about players/systems. I will say that it's hard to have these types of discussions. People so much want to run and use the first and second level of stats (first level - raw (i.e points/rebounds/assists; second level TS%/PER,etc) and I feel that basketball isn't there yet. There is still a lot of combination of needs to happen (i.e your analysis of Korver's impact).

Your post was a joy to read and the next level(s) you went to was incredibly fascinating and it's stuff that we do not see a lot of.

[Dr Spaceman]

This is where basketball and baseball analytics (the father of sports adopting analytics) have one major difference. In baseball we can say one hitter is better than the other because of certain stats such as OBP, OPS, etc. In pitching we can say one pitcher is better than the other because of a lower FIP or some other stat. In basketball, because it's five man units, it's impossible (literally) to quantify one player and then openly compare him to another player with different skill sets being asked to do different things.

In basketball, the best teams maximize the system they have at that time to fit their talents. The Hawks (and the Spurs before them) are examples of taking players that have some minor flaw but creating a system where that flaw isn't able to be extracted. So when someone tells me that Korver has a higher impact than Butler, I realize they've fallen into a fallacy of believing that stats CAN tell the whole story and in the NBA that simply isn't true. Sure Korver does things with shooting that are high valuable and create value to his team; however, that doesn't account for the fact that Butler has a different skill set. So while Atlanta wouldn't run Jimmy off screens, they could if they wanted have two viable ball handlers and create a system that plays inside out with cutters and post players taking advantage and running multiple P*R actions, similar to what San Antonio did in the mid 2000s when they were winning. And we haven't even gotten into the different things Atlanta could do defensively with Butler or the different ways they can play in transition.

And by the way I'm not saying that Butler IS more impactful than Korver or vice versa but rather the idea that you can quantify one's impact and compare it against another player is practically impossible. Both have impacts that are drastically different; without literally placing Butler on the Hawks and Korver on the Bulls for a sizable sample, we'll simply never know. Because at the end of the day, analytics tell me how much better Korver makes the Hawks and how much better Jimmy makes the Bulls but that doesn't mean you can pull them out and say "well higher RPM/VORP means better player". I think some (not saying you) are hoping that one day basketball gets to the point of baseball where basically a guy with a spreadsheet could tell you the best players in the league but that's simply impossible in the NBA.

Good post . Thanks for sharing.

Part of why I wanted to do this is because I explicitly don't believe that basketball is quantifiable like baseball. My desire is to teach people the uses, but also just as importantly the limitation of these stats.

My taking Korver over Butler is a judgement call. That's the beauty of basketball and why it's different.

Statistics are evidence. Watching games and film is also evidence. What you're doing in talking about skillsets and strategies is even more evidence. What I want to do is teach people how to use and balance this evidence in their judgements. That's analytics, it's not just stats.

Hey, if that sounded attacking at all that wasn't my intention, I just wanted to focus on one part of what you said and add how I feel analytics are important to today's basketball and to how we talk about players/systems. I will say that it's hard to have these types of discussions. People so much want to run and use the first and second level of stats (first level - raw (i.e points/rebounds/assists; second level TS%/PER,etc) and I feel that basketball isn't there yet. There is still a lot of combination of needs to happen (i.e your analysis of Korver's impact).

Your post was a joy to read and the next level(s) you went to was incredibly fascinating and it's stuff that we do not see a lot of.

Not at all, I wasn't offended in the slightest.

In general, I see a tendency for people to take one look at things like RAPM, see someone like Amir Johnson at the top, and say, "well, this isn't for me". The type of stats that are being developed at the moment can be very dense and complicated and I know people have questions about meaning and stuff like that. I think it's evidence of how far we've come that you and I can have a legitimate discussion about who is doing more between these two, when in the past it would seem ridiculous to compare a guy who scores 12 to a guy who scores 20.

So much of analytics is knowing how to use the stats effectively, and how to integrate that with the eye test.

As you and Coldfish have pointed out, the joy of basketball is in determining the "why" of a guy's impact. But the stats that are coming around are allowing us to explore those questions more deeply than we ever have before. People are opening up more and more to the idea of using deeper analytics (these are the real advanced stats, PER as you mentioned is barely a step above points per game). The important thing here is deepening the conversation. If RAPM or WOWY or everything else I'm using become standard it will allow comparisons to be so much more meaningful. If we can take it as a given that Korver's scoreboard impact is bigger, then this info is not only useful in the context of Butler vs. Korver, but we can also glean certain things about what skills are useful in the NBA and how to evaluate players that do completely opposite things.

I'm a basketball junkie, and this stuff just feeds my obsession even more.

EDIT: To add, I also like information parity, and a lot of this stuff are not the type of things basketball fans are typically aware of. If we can get people on the same page re: these stats, basketball conversations can become much more insightful.

[Hummm]

This one grouping from RAPM tell me it needs work.

Chris Bosh 0.8 0.6 1.4 4187
Carmelo Anthony 2.5 -1.1 1.4 4014
Kirk Hinrich -0.6 2.0 1.4 3685
DeMarcus Cousins -0.3 1.6 1.4 3657

These 4 players had the same RAPM of 1.4. So, Hinrich is equal to the other three?????

[Dr Spaceman]

This one grouping from RAPM tell me it needs work.

Chris Bosh 0.8 0.6 1.4 4187
Carmelo Anthony 2.5 -1.1 1.4 4014
Kirk Hinrich -0.6 2.0 1.4 3685
DeMarcus Cousins -0.3 1.6 1.4 3657

These 4 players had the same RAPM of 1.4. So, Hinrich is equal to the other three?????

Nope. RAPM is not an absolute player value. It's saying that Hinrich, in his role and minutes, is affecting the game in a positive way. Similarly, the other 3 are doing it but in a much bigger role and in many more minutes. Unfortunately RAPM leaves it up to you to make these adjustments. It is not a player ranking system, it's supposed to give you a basis for further thought.

[Leslie Forman]

Not so long ago, players were evaluated heavily just by raw stats. Points, rebounds and assists. The end result was that guys who were relentless chuckers or awful defenders had much better reputations than their contribution justified. When analytics came around, I thought they were awesome. They almost invariably lined up with my "eye test" and they were a great tool to argue with the old school type people still using points, rebounds and assists.

Analytics moved fast though. The old school people became converts rather quickly. The other day Aaron Brooks had 22 points on 23 shots while playing awful defense. A decade ago people would have thought he was the star of the game but that is basically gone. Its to the point where analytics are just confirming what most people see and they almost have little value in a conversation.

I don't think that's true. If anything back then being a "winner," whatever the hell that means, was even more valued than it is now. There were tons of guys who put up huge stats but weren't seen as "winners."

I mean, BIll Russell beat out a man who averaged 50 points a game for MVP. How in the holy hell does that happen?

[dice]

Not so long ago, players were evaluated heavily just by raw stats. Points, rebounds and assists. The end result was that guys who were relentless chuckers or awful defenders had much better reputations than their contribution justified. When analytics came around, I thought they were awesome. They almost invariably lined up with my "eye test" and they were a great tool to argue with the old school type people still using points, rebounds and assists.

Analytics moved fast though. The old school people became converts rather quickly. The other day Aaron Brooks had 22 points on 23 shots while playing awful defense. A decade ago people would have thought he was the star of the game but that is basically gone. Its to the point where analytics are just confirming what most people see and they almost have little value in a conversation.

I don't think that's true. If anything back then being a "winner," whatever the hell that means, was even more valued than it is now. There were tons of guys who put up huge stats but weren't seen as "winners."

I mean, BIll Russell beat out a man who averaged 50 points a game for MVP. How in the holy hell does that happen?

scoring 50 isn't so hard when you play OVER 48 minutes a game and average 40 FGA and 17 FTA

[dice]

Player Tracking Plus Minus

A new development in 2015, this is the first statistical plus minus attempt to use player tracking data in place of the box score. Basically RAPM is "built on" by the player tracking data mentioned above.

https://docs.google.com/spreadsheets/d/1GtCDQw94kpcOw_kPhyH8F5cIjPT3QTsOGqvrX_hMCo8/edit?pli=1#gid=0

http://counting-the-baskets.typepad.com/my-blog/2014/09/introducing-player-tracking-plus-minus.html

Why is this a big deal? As illustrated here: http://apbr.org/metrics/viewtopic.php?f=2&t=8633&start=210#p23003 it is currently outperforming every single other type of metric by a gigantic margin. Nothing else comes close to being able to predict wins. If this continues, we could have by far the best all in one metric ever created in our hands.

if a monkey wins a stock picking contest that doesn't mean she's a market wizard. gonna need a lot more than one season of accurate win predictions

[Dr Spaceman]

Player Tracking Plus Minus

A new development in 2015, this is the first statistical plus minus attempt to use player tracking data in place of the box score. Basically RAPM is "built on" by the player tracking data mentioned above.

https://docs.google.com/spreadsheets/d/1GtCDQw94kpcOw_kPhyH8F5cIjPT3QTsOGqvrX_hMCo8/edit?pli=1#gid=0

http://counting-the-baskets.typepad.com/my-blog/2014/09/introducing-player-tracking-plus-minus.html

Why is this a big deal? As illustrated here: http://apbr.org/metrics/viewtopic.php?f=2&t=8633&start=210#p23003 it is currently outperforming every single other type of metric by a gigantic margin. Nothing else comes close to being able to predict wins. If this continues, we could have by far the best all in one metric ever created in our hands.

if a monkey wins a stock picking contest that doesn't mean she's a market wizard. gonna need a lot more than one season of accurate win predictions

Well yeah. I'd definitely preach hesitance with that. There's plenty of reason to be excited though, since this is a stat building on a previous stat which is lauded for its predictive validity.

[chitowndish]

Thanks for starting this thread I think it's an amazing idea. I was looking at that RAPM 2015 spreadsheet and I do think it passes the sniff test for a lot of bulls players. Butler really hasn't been having nearly the defensive effect he usually does but has really picked it up on the offensive end. Rose fell off a cliff defensively and is treading water offensively. Hinrich is still kind of surprisingly a net positive with his d. To me that's a little surprising and shows he may get a little more hate than is warranted (I'm honestly guilty of this as well) Interesting how Curry has such a similar impact to LeBron and how LeBron has given up on D a bit. Maybe Curry's comments weren't that crazy after all. Also it became clear how much Korver improved on D under Thibs so not surprising he is a positive on that end he ended up playing really smart team D on that end and is kind of underrated for it. Same with Dunleavy too.

To me it always seems like +- is caught between two opposing forces which is quality of the people you are playing with and the quality of the person playing behind you. It also seems to me to limit one of those you introduce the other (could be wrong on this though not really a mathematician). But for example the best way to remove strength of your lineup would be to compare a lineup with a player to that same lineup without the player which solves the first problem but really strengthens the issue of player vs backup difference. Are both issues kind of resolved by the sheer number of lineups that occur or do the plus minus stats optimize in a particular direction depending on their purpose? Or am I maybe wrong that this issue is significant?

Thanks for the insight I really appreciate it!


Sent from my iPhone using RealGM Forums

[Ice Man]

Appreciate the thread. I'll take a longer look at Player Tracking Plus Minus when I get the time. At first glance, the output looked good.

[Dr Spaceman]

Should've included this in my OP, but check out GotBuckets' new FFAPM (Four Factor Adjusted Plus Minus).

Some background if you're interested:

Spoiler:

http://www.basketball-reference.com/about/factors.html

Anyone taken a look at Got Buckets' FourFactor APM? Its multiyear, they have data for the 2015 season (that is currently using data from Feb 2013-Feb 2015), but only in offensive/defensive splits. It'd be good to compare to previous seasons (they go back to 2005).

Offense

Defense

Here is how anyone that got an MVP top 10 vote last year is doing according to this stat and I listed it from largest to smallest

Chris Paul = 9.07/3.07 = 12.14
LeBron James = 9.07/2.14 = 11.21
Stephen Curry = 12.93/-2.88 = 10.05
James Harden = 11.64/-2.29 = 9.35
Kevin Durant = 5.80/0.56 = 6.36
Lamarcus Aldridge = 2.64/1.95 = 4.59
Joakim Noah = -4.48/5.68 = 1.20
Blake Griffin = 0.10/-0.64 = -0.54
Kevin Love = -0.45/-0.78 = -1.23
Al Jefferson = -4.25/-1.61 = -5.86

Also is their a bias when it comes to big men on offense because both gasols, demarcus cousins are all in the negative, blake is barely above zero and LaMarcus is the first true big man ( PF, PF/C or C to be listed and he's only at 2.64, then Davis who is a great offensive player on a good offensive team is only at 2.08??? )

Frankly I'd just ignore the big number they've amalgamated there, it doesn't seem to say how it's weighted and the defensive version is particularly poor.

Here's what's important about this stat: it's a regression on Dean Oliver's four factors, so quite literally it's attempt to gauge "rebounding impact", "eFG impact" "TOV impact" and "FTR impact". In the same way RAPM is meant to quantify "scoreboard impact".

So basically look here if you want to see "who increases their team's rebounding rate the most when entring the game?"

And here's where to find the stat: http://www.gotbuckets.com/2015-ffapm-offense/

Essentially what this is is a regression on Dean Oliver's Four Factors to determine the effect a given player has on any one of them. The players are ranked in terms of percentile (ie. a rating of 99 means that player is better than 99% of NBA players at this). so this is a good way to determine "rebounding impact", "scoring efficiency impact", and it does for both offense and defense.

Basically, if you're interested on how a team's REB%, TOV%, eFG%, or FTR changes when a given player enters the game look here.

[musiqsoulchild]

This one grouping from RAPM tell me it needs work.

Chris Bosh 0.8 0.6 1.4 4187
Carmelo Anthony 2.5 -1.1 1.4 4014
Kirk Hinrich -0.6 2.0 1.4 3685
DeMarcus Cousins -0.3 1.6 1.4 3657

These 4 players had the same RAPM of 1.4. So, Hinrich is equal to the other three?????

To add to what Spaceman just said, this also tells you that of the 4 players in the bunch.....Carmelo has the most impact on offense and Kirk the least. And vice versa on defense.

When you think about a minute for minute comparison , that conclusion is pretty reliable. A classic example of numbers backing the actual observations on court.

There's also other nebulous questions you can ask based on that particular grouping you chose:
1) Does said player make the entire team better?
2) Which of the 3 star players in the above grouping has the greatest impact on the game, etc?

For example, in question #2, I would look at those numbers and say that Bosh makes the most impact to his team. And Anthony the least. The range of their off/def RAPM numbers is what I used to eyeball that.

[Dr Spaceman]

Thanks for starting this thread I think it's an amazing idea. I was looking at that RAPM 2015 spreadsheet and I do think it passes the sniff test for a lot of bulls players. Butler really hasn't been having nearly the defensive effect he usually does but has really picked it up on the offensive end. Rose fell off a cliff defensively and is treading water offensively. Hinrich is still kind of surprisingly a net positive with his d. To me that's a little surprising and shows he may get a little more hate than is warranted (I'm honestly guilty of this as well) Interesting how Curry has such a similar impact to LeBron and how LeBron has given up on D a bit. Maybe Curry's comments weren't that crazy after all. Also it became clear how much Korver improved on D under Thibs so not surprising he is a positive on that end he ended up playing really smart team D on that end and is kind of underrated for it. Same with Dunleavy too.

Yeah the reason I fell in love with RAPM in the first place was that it was basically quantifying the things I was seeing on film. Most RAPM evangelists I talk to love it for those same reasons.

To me it always seems like +- is caught between two opposing forces which is quality of the people you are playing with and the quality of the person playing behind you. It also seems to me to limit one of those you introduce the other (could be wrong on this though not really a mathematician). But for example the best way to remove strength of your lineup would be to compare a lineup with a player to that same lineup without the player which solves the first problem but really strengthens the issue of player vs backup difference. Are both issues kind of resolved by the sheer number of lineups that occur or do the plus minus stats optimize in a particular direction depending on their purpose? Or am I maybe wrong that this issue is significant?

I don't know how comfortable you are with deeper mathematics and statistical theory, but the theory of ridge regression is basically an answer to the concerns raised in your post. Essentially, we are determining the relationship between two variables here, player x in-game (independent variable) and scoring margin (dependent variable). The problem, like you said, is that there are so many confounding variables that it's very very difficult to isolate a single player as the cause of a change in scoring margin.

This is where regression analysis comes in. It is a peer-reviewed statistical method to "fix" all of the confounding variables to a certain value so that the relationship between the two variables of interest can be analyzed. So quite literally this can be seen as an "experiment" where we place every player in exactly the same context (teammates, opponent, etc.), manipulate the presence of the player in question, and determine what we can expect from them given that neutral context.

There are other methods of doing this, of course, and if you take a look at ElGee's work I linked above you can see how he went about controlling for these confounds. No method is perfect, but generally when these types of stats "agree" with each other ("concurrent validity") or display the same results over a long period of time we can be reasonably sure the results are accurate.

It's important to understand the reasoning behind these stats as well. For a long time we relied on the box score, which is not an accurate measure of a player's "impact". So we can call the box score approach "invalid". However, the box score has a high degree of reliability, ie. the results are very dependable and predict future results very well. By contrast, RAPM has a high degree of validity, meaning it does a far better job of measuring what we want to measure (a player's impact), but it severely lacks reliability compared to the box score.

This can all be summed up in one concise sentence: If you're looking to asses a player's impact, RAPM has a much greater chance of giving you the correct answer than the box score does, but it also has a greater chance of screwing up worse than the box score ever could. This is why we use prior results to inform current ones, so the stat doesn't get as fooled by crazy results.

Of course, neither stat can determine how "good" a player is, that's where your own brain comes in

I'd be willing to go deeper into the mathematical mechanics of the stat as well if you're interested.

Thanks for the insight I really appreciate it!

Always!

[Ice Man]

My take on Kyle Korver:

Spoiler:

I meant do you believe that Korver, e.g., really has had much more positive impact on the court for Atlanta than Butler has had for the Bulls?

There's a reason I have a picture of Korver as my avatar

You may not realize it, but this is one of the most important questions you could've asked in your quest to understand impact statistics. This is generally the part of the conversation where people tend to look at me sideways or even outright dismiss me, so I really hope you'll read this with an open mind.

Think about this: on offense, a basketball player is always in one of two conditions. Either he has the ball, or he doesn't. A player can contribute by doing good things with the ball (this is what's captured by the box score and what most people focus on when watching games) or he can do good things without the ball (this is something that isn't traditionally measured by any of our counting statistics and something most people don''t notice).

Now with that in mind, consider Korver's shooting. Obviously, people see that he scores 12 ppg and question how this guy can be seriously considered an impact player. But consider a couple things: among players who have equalled Korver's 3 point shooting volume in history, the highest percentage ever recorded was 45%. Korver is currently shooting 49.6%. He's bombing 3s at a rate equivalent to some of the highest in history, and he's hitting them at a percentage no one has ever come close to. On Catch & Shoot plays, per NBA.com, Korver is averaging over 1.3 points per possession. The best offense in NBA history scored about 1.15 points per possession.

Now another thing Korver has mastered is being a brutally quick decision maker. He catches and either shoots if he's open or finds an open teammate to pass to. This means that failed attempts to get him the ball cost very little shot clock time, and he's not taking bad shots when he does possess the ball. This is the most underrated aspect of what made guys like Bird and Jordan so ridiculous on offense. It was always catch, read, attack or pass within the blink of an eye. Korea does this as well as anyone, and the result is he doesn't waste possessions. His teammates who are better at isolating and pick and roll have more shot clock to work with.

And another thing in Korver's favor: right now he leads the league in True Shooting Percentage. What that means in practice is that a shot attempt by Korver on average results in more points than a shot attempt by any other player in the NBA. This is a big, big deal.

Okay so we've established what Korver does with the ball that makes him so special, but what about without the ball? Well, given that he's the best in the league when he actually does decide to shoot, any reasonable opposing coach is going to do their best to keep the ball out of his hands. And this is the secret to the Hawks offensive success. It's almost a joke how far opposing teams will go to stop Korver from shooting. I'm talking doubles at the 3 point line, switching every screen, I've even seen multiple times this year opposing defenses concede an open dunk just to avoid Korver being open. It's ridiculous.

And Korver does not behave like a normal shooter. Watch him during a Hawks game sometime, he's constantly moving, sprinting in circles around the arc, making decisive dives to the corner. He's the biggest threat on the Hawks, and all eyes are on him all the time. Defenses literally treat him like he's LeBron.

We imagine the typical low volume 3 point shooter as a guy who sits in the corner and shoots when he's passed to. But this is not the case with Kyle. The Hawks will run him around screen after screen, watch the defense shatter, and then attack the craters he leaves behind. They are so brutally effective at using Korver's gravity to set up 2-on-1s and mismatch isolations. Korea doesn't get credited with an assist on these plays, but he deserves it.

And think about what you'd do if you're defending a pick-and-roll with Korver lurking on the weak side. You absolutely cannot help off of him. In Thibodeau's defensive scheme, the weak side corner help man needs to drop into the paint and bump the roll man, impeding his way to the basket. But if that defender is guarding Korver, you can't use that guy to help on the pick and roll at all. Watch the Hawks run pick and roll, and it honestly looks like you could drive a semi through the lane at times.

So yeah, Korver is having tangible, transcendent impact every single possession, and he never even has to touch the ball to do so.

So big picture, what does this mean? I don't think Korver is a better basketball player than Jimmy Butler. Butler has much better defensive told, a more versatile skill set, and can be a high-usage guy in a way that Korver can't. However, Jimmy Butler is not having as much impact on the basketball court as Korver is right now. Very few players are.

And this gets a little philosophical, but follow me here: we tend to think of basketball players in terms of how well they could do without quality talent around them. But to me this is a fallacy, because if the goal is to win a championship, I want players that play really well when they have other talented players around them, even if they need other talent to do their thing effectively.

A real world example: The Knicks would be far, far worse with Korver in place of Carmelo. This is a given, since Carmelo is much better suited to carrying a team like that. But flip things around: do you honestly believe the Hawks would be better with Carmelo in place of Korver? Because I legitimately don't. I think the Hawks would be worse with Carmelo.

Another real-world example: You have the option of putting any shooting guard in the league on the Cavaliers. do you pick anyone other than Korver? The guy is going to make life so much easier for the other offensive stars, and when he does use possessions, he's going to use them better than even the other stars would!

Now what you'll take issue with is Korver's context-dependency. And that's totally fair, and for a lot of teams they wouldn't hesitate to take a litany of other guys over Korver. But think about this: is it more valuable to make a bad team mediocre, or to be able to make a great team championship-worthy? Because the latter is what Korver does. And I'll take that over the Carmelo's of the world any day.

Wow that was long. Anyway this is why I'm such a huge advocate of film study. Once you accept RAPM as being a valid indicator of "impact" you start looking at basketball in a different way, and start to understand the unseen impact of guys like Korver. Other big ones are Steve Nash, Bill Walton, David Robinson, etc.

This post should be required reading. It would fix three major misconceptions on this board -

1) That use of the statistics means not watching the game. (The Barkley Fallacy.) Quite the reverse. Statistics provide information that make for a better watcher. And watching makes for better statistics. The two items work together, in a feedback loop.

2) That several pretty good skills beat a few skills. (The Marco Fallacy.) There are many, many wings who beat Korver on a skills checklist. For example, Belinelli is bigger, faster, a better dribbler, better creator, more agile. Snell is quicker, longer, a better leaper, better lateral movement, better finisher. But Korver is REALLY REALLY good at his skills of shooting, moving without the ball, and decision making, and that makes him the better player.

3) That many players can become Korver 2.0 (The McDermott Fallacy.) Nobody has ever done quite what Korver has done the past 3 years. There are many, many 6' 7" guys with average speed who can shoot. There is only one Korver. To blithely project that a rookie who seems vaguely like a young Kyle can achieve what Kyle did is to greatly underestimate just how special Korver's development has been.