Saturday, May 01, 2010

Doesn't This World Need Another Stat?

In watching another Mavericks playoff disaster, I have had this twitch inside me to pursue something that has troubled me for almost a decade of watching Mavericks Basketball.

If you have heard the show at all after Mavericks playoff games going back to the Michael Finley-Don Nelson era, you certainly have heard me say, "The Mavs cannot win by trading 22-footers for 2-footers." What that means to me is simply this: The Dirk Mavs seem to design offenses that free Dirk and his mates up for open perimeter looks. This has served them well many nights in the last decade, but as is the case with any perimeter-based offense, there are just nights where you cannot "throw it in the ocean".

On the other end of the court - perhaps, again partly because you have a 7-foot power forward named Dirk - the Mavericks are usually not able to force playoff opponents to settle for jump shots in the 4th Quarter. Too many times, whether it was Mike Bibby, Tony Parker, Chris Paul, Dwyane Wade, Baron Davis, Manu Ginobili or someone else, the penetration into the paint doomed the Mavs when the game turned into a half court match of wits.

These playoff games often come down to a contest of execution in the half-court game. On offense, The Mavericks' best chance was the high pick and roll of Nash/Dirk or Terry/Dirk. On Defense, the inability to force contested jump shots, put the pressure on the offense to never lose serve. To make matters worse, the offense would not draw many fouls. The defense would be under constant stress as slashers ate the Mavs for lunch.

And that is how you lose 16 of 18 road playoff games. Simplistic explanation, but sometimes simple is all you need to explain basketball. You have one of the truly special players of this generation, and yet nobody has been able to figure out how to win in the playoffs with Dirk as your centerpiece.

And that leads me to this project. First, understand that although this project has been in my head for several years, this is the first time I have actually sat down and tried to figure it all out on paper. There are some flaws in the logic that I need to sort through (perhaps with your help). Second, my idea seems so simple that there is a good chance someone else has already thought of it. I don't want you to think that I believe I am the first person to ever think of this, because there are so many smart people following this game that surely this has been out there before. I just haven't seen it anywhere, and that is why I want to see what you think.

My idea is this: Rather than making the claim that the Mavericks "settle" for perimeter shots consistently and claim that the opponent does not, maybe there is a way to actually prove it. I can never measure defensive intensity. I cannot measure desire. I cannot measure the harshness of officiating. BUT, I can measure how far the Mavs and their opponents are from the basket when they shoot the basketball.

The NBA keeps a full play-by-play of every game in which they list the rough estimate of the distance from the basket for every shot attempted in each game. What if I merely tabulated all of the shots and found the average distance for that night's shots? It sounds so simple, and yet it would seem to reveal whether a team was "settling" or "taking it to the hoop".

So, I did. WIth tons of tabulation help from TC, below are the findings of the entire series between San Antonio and Dallas; Until we figure out a real catchy name for this simple metric, let's call it "Average Distance per Shot", or, "ADS".

Mavs ADSGameSpurs ADS
13.8Game 113.6
14.5Game 214.7
13.0Game 310.3
13.0Game 413.7
15.0Game 514.2
13.4Game 610.5

HTML Tables



Now, you may look at that chart and aside from a few observations the numbers might all seem very close. A few things like, Wow, according to the Spurs ADS in Games 3 and 6, in particular, they were getting a ton of shots from a very close range. However, I am not happy with the findings, because in Game 5, the Mavericks had their worst ADS of the series (15.0), but I think most of us feel that was the game where they took the ball to the basket the best. So why is their ADS so bad? Because late in the game when it was decided, the bench players started launching 23 footers to kill the average.

But, when I broke it down by Quarter, you will see that in the 1st Half, the Mavericks dominated the ADS stat on both ends of the court. They were shooting from close range more often, and forcing the Spurs further away from the basket, on the way to a 53-46 halftime lead. Then, up 82-64 entering the 4th, the game was basically garbage time for the rest of the contest. Look below at Game 5 by Quarter:

Mavs ADSGame 5Spurs ADS
12.91st Q14.6
12.62nd Q15.0
14.53rd Q12.2
18.64th Q15.3
15.0Game 514.2

HTML Tables



I think we are starting to see that a full game does not tell us a whole lot of information, but, when you start to break it down by Quarter, we see where the trends in the game were shown on the scoreboard.

Remember Game 3 in San Antonio? The game made famous by Manu's broken nose, JJ Barea's 32 minute stint, Danny Crawford's record continuing in Mavs games, and the Mavs fade in the 4th tells us plenty:

Mavs ADSGame 3Spurs ADS
10.11st Q11.6
14.52nd Q10.1
13.43rd Q9.3
14.04th Q10.5
13.0Game 310.3

HTML Tables



This game showed us that it wasn't just the Mavs offense. Actually, the offense is about as good as get it gets for the Mavs with an average shot distance of 13.0 for a full game is the best they did in this series. But, it was the defense in this game (and if you saw this game, you will remember that Manu and Parker got in the paint over and over and over again) driving the Spurs' ADS down into single digits! For the game, the Spurs ADS was barely above 10 feet (10.3) which is the lowest number of the entire series.

Look at each quarter of Game 3; In every Quarter but the 1st, the Spurs are standing a full 4 feet closer on EVERY shot! This is the fundamental problem that the Mavs are trying to fight every year at playoff time. When teams are "quality opponents", they generally have the ability to play their most vicious defense at key moments. If one team can make another team stand 4 feet further on every single shot, over the course of 50 shots, that will be next to impossible to overcome, it seems.

Now, look at Game 6:

Mavs ADSGame 6Spurs ADS
14.21st Q10.0
11.02nd Q8.6
13.23rd Q10.8
15.84th Q12.4
13.4Game 610.5

HTML Tables



In the elimination game, with both teams knowing how well the Mavericks opened Game 5 by attacking the rim and owning the paint, the Spurs decided to take it back in Game 6. Every single Quarter, the Mavs were forced much further away from the basket than the Spurs. 3 feet on average. When the Mavs took the ball to the basket some, the Spurs did more. And as the game continued, the Mavs creeped back because they hit some long shots, but no team can sustain it. You can get hot, but you can never stay hot. If you can stay hot for a game, you cannot sustain it for a series. Trading long shots for short shots is the formula for disaster.

Again, the average distance of every shot a player or a team takes seems like a rather easy way to determine how aggresive the team is attempting to be. There are surely flaws in it. Here are a few:

* If the sample is too big, it will not recognize the intensity drop of "garbage time". Not every position is against the same intensity. A possession in a tie game with 30 seconds to play versus a possession against a 25 point lead cannot be compared.

* Shots that result in fouls are not included as "Field Goals Attempted". Obviously, if a player is fouled trying to dunk, he gets rewarded with 2 free throws, but would not have any bearing on his "ADS". This is a problem.

* Desperation heaves from 75-feet would really throw off the sample. Therefore, I do not include desperation heaves in the statistics. I could not think of any other solution to that.

But, beyond that, I feel like this is so basic and yet so telling. It tells us that every team shoots from the perimeter, but some teams shoot much more from the perimeter. It tells us a team like the Spurs - due to penetration and the presence of a post scorer - shoot from very close to the basket quite often. And, of course, it tells us that the Mavs are trying to win a game in which they rely on 20-footers falling in crucial moments. In all of these games, as the game got later (And intensity is turned up) the Mavs had to keep backing up to get shots. Always a bad trend because sometimes those shots fall, sometimes, they get eliminated before May 1st.

Like I said, someone may already keep this stat. Or, maybe it is so simple it has no practical application to NBA minds. But, it would seem to me that if you take 10 footers instead of 20, you have a better chance of making them. Note: Of course, a wide open 20 footer is easier than a contested 10 footer, but over the course of a sample, I believe this evens out quite a bit. Uncontested lay-ups are way easier than uncontested 3's. Contested dunks are easier than contested 3's, and so on.

If the Mavs are more aggressive, their ADS drops from 15.0 to 12.0. When they get more passive it raises from 12.5 to 17.0. And we can see that when we watch the game. We also know that Jason Kidd is nothing but a perimeter shooter anymore. We see it, but now we know that his ADS in Game 6 was "23.0", but Roddy Beaubois in Game 6 had a shockingly low "10.3". Yes, Roddy took a few 3's, but he also took quite a few "1-footers" to show he was aggressively taking it to the rim.

In a future post, I want to look at every player on the roster from an "ADS" standpoint, as well as stars in the league like Kobe Bryant and Lebron James to see what all of this means.

But, for now, let me know what you think.

27 comments:

rncantu said...

I like where you're going with this but I think this tells us something different.

I think Dirk drives to the basket a decent amount. He certainly drives more than he used to. I think Butler goes baseline and gets to the rim a decent amount as well. Marion only usually takes shots from around the painted area which is a good thing. Yes, Jason Kidd is only a wide open drive and kick 3-point shooter. Jason Terry's game is built around the ten to fifteen foot shot and he's still decent at that. What I'm trying to say is that the Mavs drive the ball as much as other teams they play. What they don't do is throw the ball into the post and let a center score for them. Look at the teams still alive. Duncan, Gasol, Bynum, Stoudemire, and Carlos Boozer are all primarily back to the basket guys.

The Mavs only low-post scorer is Brendan Haywood but I guess Carlisle forgot that he said Haywood "changes the geometry of the game". Instead he goes with one of the worst low-post players in Dampier. I'm betting if you checked the "ADS" from the 13 game winning streak (when Damp was hurt) you'll find it to be much lower. Haywood was close to averaging a double-double during the streak.

Unknown said...

Great stat. What about ads-made vs ads-missed?

Doctorjorts said...

I agree a lot with what rcantu said, though I think he's also making your point for you... the Mavs' scorers are jump shooters for the most part. What you need is just one guy who can get a close shot, whether through penetration or low-post moves. Beaubois was clearly the former in game 6, and if he can continue his upward trend into next season I think he could be a huge weapon. It's a piece the Mavs haven't had since Harris did that for them in '05-'06.

That said, if it's Dirk that's taking shots from 15 feet out, I'm completely fine with that. He's generally such a solid shooter that I can live and die by his jump shooting. I'm not willing live and die by Butler's, Terry's, Marion's, Barrea's and (to a slightly lesser extent) Kidd's jump shooting. I don't know what Butler's game was like in Washington, but he and Terry absolutely must be the penetration threat to complement Dirk's outside game when Dirk is on the floor.

It's the willingness of players other than Dirk to settle for jumpers that is really the problem we're seeing in the ADS trends.

Dylan said...

I think you've got the right start on this.

And I think you're right to take desperation heaves out of the equation.

The next step would be to create a "Situational ADS" by adding weight to point differentials during the game.

You'd have to play with it to get the right multiplyer, but say 3x when game is score difference between the two teams is 0-4, 2x 5-9, 1x 10-14 and .5x 15+. It would invert the way you look at it, so the higher the number the better, but it would help to balance out trashtime baskets.

I haven't run the numbers this way, but it might be interesting to see what showed up with these multiplyers.

the inflationista. said...

bob - beautiful work you've done -

norm uses a stat in football called yards per point -

like chris said, if you were to weight only buckets that fell by the distance from which they were taken you'd see what the functional range is for a team -

in basketball, you might consider average distance per 2 points -

by using 2 points instead of 1 you a) weight the distance more closely to reality; b) show teams that rely on field goals from inside the arc a better distance bias -

your work is almost inspirational, i kind of want to shed a sports tear for your obvious appreciation for the game -

Unknown said...

Good stuff Sturm. I think a Win-Loss record when a team wins the ADS would really sell the theory that you win a majority of the time when you attack the rim.

Unknown said...

Yeah, this is great. Need to add stnd deviation and plot over time. Layer point diff, who is on the floor ... do it in finer increments ... and _animate_ it!

Where is the source data?

Jim said...

Sturminated. Good work.

I wonder if median distance per shot (MDS) would offer a way to cut out the outliers. There are problems with both mean and median, and I'm no statistics expert. But it seems that you've got enough data that it would be worth looking at the MDS.

Maybe there's a way (this is way above my pay grade) of combining MDS and ADS?

JB in Detroit.

Prophetic Rain said...

This stat brought back memories of the Kings series in 04 where bibby and jackson destroyed the mavericks with their penetration.
Althought in looking at the mavericks over the years compared to every team that one the championship in the otts, I see that the Mavericks only player that can score consistently in a half court offense is Dirk. He can create his own shot better than any other player that has been on the roster, with exception of Caron butler now. Every championship team in the otts? SPURS, LAKERS, PISTONS, & CELTICS had a consistent half court offense that let them move the ball creating quality shots. What Roddy B and Butler can do is give slashing wings that can shoot perimeter shots. The prior rosters had Howard, Finley, Nash, Harris with Dirk. Harris was the only slasher of that group while in Dallas. Kobe, Lebron, & DWade take it to the hole consistently and dictate to their team what needs to happen. This is how Cleveland erased a dominate 14 point lead from Boston and took the game over. Lebron slashed and passed to guys with open looks that slashed themselves, ala Mo Williams dunking on Paul Pierce. Thank you sports sturm I'm all gigged up about bball again after dealing with the series defeat.

Just Me said...

One series is a poor sample size. You'll need to research every playoff series of the last 5-10 years if you want to show that there is a correlation (much less a causation) between ADS and wins. And don't forget to run a regression or two.

Tell TC to have fun!

alwais on said...

great stats. i love the numbers. i think what this does bring to light, is that they are not "settling" for jump shots. I think with the mavs, i think their offense is to get jump shots. so settling may be the wrong word, i think that the mavs offense wants jump shots, they are a team of jump shooters. even Dirk is known for his uncanny talent as a 7 footer who can knock down the outside shot, that is what makes him so difficult to defend. your amazing work, proves that the mavs are a team that looks for the jump shot and takes it when open rather than looking to get into the paint.

Unknown said...

While I agree with your overall premise and agree the stats support the theory, I think you miss one very important element. How many jump shots made are the result of pentration or a double team post. Having watched the Spurs closely for years, I can tell you the fact they have guys near the top of the league in 3 pt shooting is not because they are great shooters. Think about all of Bowen's corner 3s. They come from a pass out of a double team on Duncan or kick out from a pentrating guard after the defense collapses. So, I'm not sure shot distance tells the whole story.

mitthrawnuruodo said...

If you regress the success of a shot (hit or miss) against the distance of the shot, you could figure out the exact impact one foot (or for that matter, one inch) has on the probability of the shot falling. With a larger sample size, like an entire season or several seasons, you could regress whether the team wins against the ADS and find out exactly how much taking shots an average of one foot closer affects winning. I don't know if that's in TC's skill set, but it sounds like an econometrician's idea of fun.

MK said...

I think a flaw with this stat is that it sort of presumes a long shot is always a bad (lower percentage) shot. An open shot is a good shot in the NBA.

ben8gan said...

Interesting read, and I can see how this surely beats FT attempts (as it accounts for potential referee bias)...however, how is this superior to a time-tested stat like Points in the Paint?

Where this perhaps could be worthwhile is looking for trends when certain combinations of players are out on the floor.

Unknown said...

Instead of average, I'd like to see them broken down into categories (i.e. 0-5 ft, 5-10 ft, 10-3 pt. line, 3 pointers)

Some really good offensive teams seem to get either dunks or open 3 pointers (Orlando). That is far different than a bunch of mid-range jumpers, but it would probably look similar on the ADS.

Good work.

southwick said...

Game 6 is so telling.

Mavs losing badly in 1st. Roddy comes in the second, and starts going to the rim. Mavs make a push. Roddy stays in, and while higher we still are fairly low for the mavs.

4th quarter, Jet comes in, and the number jumps way up to 15.8.

mycubans said...

Great analysis Sports Sturm.

I agree with @Jim. Median should give a better picture of where the shots were actually coming from. It would reduce the impact of the 4th quarter time-kill heaves. While you could hand pick a few to leave out (half court shots at the buzzer for example) it would be hard to do that on a large scale and would definitely bias the results. Median does it for you - for the most part at least.

Unknown said...

There's a good reason the Mavs take long jumpers. Besides at the rim and on threes (every team's most efficient shots),the Mavs fg% from 16-23ft is higher than from inside 10 feet and from 10-15 feet.

http://hoopdata.com/teamshotlocs.aspx

I know it seems to make sense that shorter shots should be more makeable, but there is definitely a good reason that the Mavs shoot more from range - it's a better bet for points!

I'm generalizing here, but just because basketball "experts" think teams should play a certain way (shoot closer shots; long twos are for sissies), doesn't mean a team should give up what works for a closer, but less efficient shot. For the Mavs, it doesn't make sense.

Anonymous said...

A couple of things.

(1) Are fouls included in this? That is, a player who goes to the rim, gets fouled and shoots free throws ... that should count as a close-by shot attempt.

(2) I think you should omit 3-point attempts, and change the stat to Average Distance per 2-point attempt. 3-point attempts should be in a different class, as there is a certain pay-off (extra point) for the risk taken.

Dahlsim said...

Excellent stat!

You can't ignore the fact that % is higher closer to the basketball but as you said it's more than that. The chances of getting freethrows is greater for an inside shot.

That produces a double benefit by producing freethrow points while also getting the opposing players into foul trouble. Foul trouble for the other teams produces multiple benefits. Earlier penalty, tenative defense from the opponent and forcing the other teams substitution patterns.

Dahlsim said...

Even it Mavs shot better from the outside (which may happen for a short time but won't happen over the long haul) they would lose out in other areas such foul trouble and points off free throws.

Even Jump shooting teams need a strong inside game to be effective at the highest levels.

Dahlsim said...

1 other factor to consider is shots taken inside from your Big Men, front line players vs. Points from smaller Men.

Both are great esp. if you have talented slashers but a 3 foot shot from a 7 footer is generally harder to stop than a 3 foot shot from a 6 footer.

Translation, Mavs need post up inside points from Center and Power forward as well as Slash & Drive. Shaq complemented D.Wade and Kobe Bryant. Big men scoring needs to be at least a viable threat to keep defenses honest all around.

Unknown said...

Bob, there is truth in what you say and your research is, as usual, impeccable, TC help notwithstanding. I think what it really gets down to is this: One player cannot win an NBA Championship, and that’s all there is to it. Even if you have a Hall of Fame player, you still need another legitimate, high-level All-Star to be able to win. That Kobe couldn’t win without Shaq or until he got Gasol doesn’t make him any less of a Hall of Fame player. The Celtics had to get Garnett and Allen for Paul Pierce. I mean, even Michael Jordan had Scottie Pippen. This is why I get so frustrated listening to some people bashing Dirk – ten years of ultimate playoff futility are not his fault – these idiots that crow about him obviously have not looked at the numbers – he has been spectacular in the playoffs, period. But he has never had his Scottie Pippen, and that is really the bottom line through all of this. Shot selection, schemes, coaching, officiating, execution, focus, etc., they all make a difference in some way. But it just comes down to this. No Scottie Pippen.

fickle said...

I agree with others. . . there's something revolutionary here.

I'd recommend getting with a stats professor or PhD candidate at one of the local universities and having them play with the numbers to discover the best and most descriptive way to calculate and demonstrate what you're dealing with. They'd probably get an interesting article out of it, too.

Mean might be better. . . using groups might be better. . . displaying variance or standard deviation might help. . . running the regression for made shots against distance would help to explain the significance of the stat. There's something here.

Unknown said...

Dear Bob

Great idea and work of yours.
Want I would need to know is, how does the ADS from the Phoenix-Spurs series look like (Suns 3 - 0 Spurs)?
My logic tells me that the figures in this series should tells us that Phoenix ADS (compared with the Spurs ADS) is better than the Mavs ADS (compared with the Spurs ADS). But if it would look similar like the Mavs ADS, then this would tell me that a jump shooting team can win a series against a spurs-style team, but that the mavs are simply a not-so-good jump shooting team

Dahlsim said...

In reality not all jump shots are created equal.

A open jump shot that comes back from an inside offensive attack can be a much better shot in forced outside shots with guys breathing down your jersey.

Inside to outside offense gives better jumpshots.