ESPN logoTrueHoop Network
An ESPN Affiliate
Daily Thunder.com
  • Home
  • About
  • Advertise
  • Contact
  • Team
    • Salaries
    • Roster
    • Schedule & Events
  • Commenting Guidelines

Getting a little more defensive about stats

by Royce Young on February 23, 2009 at 1:13 pm 17 Comments

Last week on the B.S. Report, Bill Simmons had John Hollinger on and they talked a lot about the limitations in NBA statistics. And one reader took that as a challenge and watched the Thunder/Warriors game with a little bit more of a keen eye. Thunder fan Taylor sent this great breakdown in of Saturday night’s game (HERE and HERE are his “box score”):

I am a huge fan of stats and wish that there were more of them that were recorded.  There are many little things that players do that don’t show up in the box score. Because of this I have started to take some in depth stats of Thunder basketball games. I do this not just because of my love for stats or the Thunder, but because it helps me become a more educated viewer while watching the games and allows me to really look for things that I would normally miss.

Attached were some footnotes explaining each stat. Assists have five categories: Key (pass led to a bucket in the painted area), Open Jumper (pass led to an open jumper), Open Three (pass led to an open three), Contested Jumper (pass led to a made jumper, but it was closely contested), Contested Three (pass led to a made three pointer, but it was close contested).

Doing this is very interesting as we can see that all six of Watson’s assists led to baskets in the key, meaning his assists were very efficient and productive (try to disregard how he played in other aspects of the game, haha). This isn’t implying that the other assists aren’t important but it is interesting to view them in these categories.

I like the idea of the next category, because I battled the same issue as a player.

The next category is labeled “Could Have Been Assist”. This is one that my dad invented in high school for me after a few frustrating games in a row where teammates missed wide open layups/jump shots after good passes. The first five categories are the same as above with a sixth added: Free-Throws.  These categories are fairly self-explanatory: if Westbrook feeds Durant under the basket and KD misses the lay up, Westbrook gets a “Could Have Been Assist – Key” in the stats. This can show how much of a play maker someone was during the game even if they aren’t showing up as official assists. The “Free-Throws” category is for any play where the recipient of a pass is fouled while shooting a missed shot. If that player makes at least one of his free-throws then the passer gets a “Could Have Been Assist – Free Throw”. The playmaker is then rewarded even though in the official box score an assist will not be given.

The main problem with CHBA is that when you’re playing a team sport, you have to rely on the other guy to finish the play. In football, you can’t give Peyton Manning a “Should Have Been Completion” because the ball went through Dallas Clark’s hands. You can’t give Vlad Guerrero an outfield assist on a bullet to the plate when one of the Flying Molinas has it bounce off his mitt. I get the logic there entirely, but in reality it doesn’t work. I do think players should get an assist if the guy is fouled on the shot and makes both free throws. The problem there is how do you award the stat if he makes 1 of 2? Half an assist? I think that’s the main complication there.

The next three categories break down the assists in three categories: Fast-Break, Half-Court, Inbound. Taylor explains that this allows us to see where players and the team are obtaining most of their assists whether in the half-court, during a fast-break or off of an inbound play.

The next four categories all generate percentages for us to evaluate.  The first is “Fast-Breaks” and then “Success”. I have since decided that these four categories will be broken down the following way: “Fast-Breaks” “Points” and “Success”. For the game against the Warriors I only have the two categories. How this works is anytime a player leads the fast-break that is reflected in the stats. If that fast-break led to a basket then it is shown as a successful fast break.  In the new system a successful fast break won’t just be from a made basket. For instance, there may be a fast break where Westbrook finds a wide open Durant trailing for a three.  If Durant misses that three, it should not be reflected in the stats that Westbrook had an unsuccessful fast-break.  He did his job and got an open look. In that scenario Westbrook would see the following in the stats: “Fast-Break” 1 “Points” 0 “Success” 1 because he did lead a successful fast-break even though points were not scored. If he then leads a fast break where he turns the ball over, followed by a fast break where he attacks the rim and scores his stats would be updated as follows: “Fast Break” 3 “Points” 1 “Success” 1. This means that out of the three times Westbrook has led the fast-break, he has been successful twice, although only producing points once.  For the Warriors game, only made baskets are recorded as being successful.

Next, Taylor looks at outlet passes. This is such an overlooked component to basketball. Against Phoenix, the Thunder was torched by Shaq and his pinpoint outlets. Taylor says, “Jeff Green did a great job last night of starting the break by making a quick, smart outlet pass. If the fast break eventually leads to a basket or open look, Jeff Green will either get a tally in the “Points” or “Success” category.”

“On Ball Screens” measure the amount of on the balls and the success rate of them. It follows the same “success” rules as above. “Crash Boards” is a category that measure when a player attacks the glass on the offensive end.

“The box score does reflect Collison’s hustle with his five offensive rebounds, but it is very informative to see that he crashed the boards (attempted to get an offensive rebound) 22 times.  It is counted when a player makes a clear attempt to grab the ball or get around the defender boxing him out.”

So I guess you could take that two ways: Nick is working hard on the glass and trying to get second chances, which is good. Or that while Nick works hard on the offensive glass, he comes up empty a lot. The key to this stat (as in all stats) is proper comparison. What’s Paul Millsap’s Crash Boards stat? What’s Dwight Howard’s rate? Obviously Taylor can’t put together that stuff, but that would make this stuff much more interesting.

Fouls Drawn and and Charges are exactly what they sound like. Now defensive stats:

“Contested Shots” is simple: did the defender get a hand up?  Was he attempting to contest the shot?  Did he close out on a shooter?  ”Altered Shots” is basically a blocked shot that is not reflected in the box score because the defender did not actually touch the ball. It is more valuable than a “Contested Shot” because you are not just putting a hand up, you are making the player change his shot in a drastic way that often leads to a miss.

“Deflections” are kept to measure how active a player is. These deflections can come in a variety of way: off passes, rebounds, loose balls, etc. Anytime a player deflects or tips the ball, whether or not it leads to a Thunder possession, that play is accounted for.
Deflections is something that I don’t understand why it isn’t kept track of. I realize not all deflections result in a steal, but they disrupt the flow of the opponent’s offensive and is a positive defense play. That just seems like an obvious stat to keep.
“Box Outs” are self-explanatory and are recorded when a players makes an attempt to box out the opposing player, not just standing there waiting for the rebound. Obviously, many rebounds are obtained without a box out but I love seeing who is putting forth the effort to put a body on someone. “Give Up Rebound” occurs when a Thunder defender is responsible for giving up an offensive rebound to the opposing team whether by being out of position or missing a box out. The Warriors only had three offensive rebounds.
“Ball Denials” are recorded to track when a player denies the ball or does not allow a pass to be made. Durant did a great job last night of not allowing the ball to be thrown into the post which is shown in the stats by his 4 ball denials. This stat can occur in the post, on the wing, etc. “Help Defense” shows us how much a player is helping on defense when his teammate get beat off the dribble or when a rotation needs to be made. The big men (Krstic, Collison) get a lot of these because they are in the paint and in most cases help out when a guard or forward is beat off the dribble. They also are the players who most often guard the picker on a pick-and-roll play and consistently have to help out. I was surprised to see how much Collison was helping out his teammates. It is mainly because of his position but it also enabled me to see how much he was hustling on defense and moving his feet.
“Charge Attempts” and “Charges Called” record how many times a Thunder defender attempts to take a charge and how often that attempt is rewarded with a charge call by the official.

I really like getting a percentage on charges attempted and charges called. It could help us distinguish the floppers and the guys that actually do a great job of getting in front and getting their feet set. That may be my favorite stat of the bunch.

So there you have it. And as Taylor notes at the bottom:

I am sure there are more stats that we can come up with but these are a few that I thought would be valuable to track.  I am not a certified official but did play basketball competively through high school.  Naturally, my stats will not be 100% (For instance, I have Westbrook down for 10 assists – box score shows 11) and many of the categories are very subjective but I still feel one can gain from looking at these stats. The Warrior game was my first attempt (watched the game recorded in case anyone thought I would even try to do this all alive) and I am sure I missed more than a few stats.
There’s flaws there for sure, but there’s also good stuff to digest. With a little more refining, some of these could be worthy stats. I’m sure coaches on the bench probably keep a pretty good track of close-outs, contested shots and box outs, but in lieu of the Shane Battier article that changed the world, it might be nice for the average fan to appreciate some of that stuff. Take it all for what it’s worth, but obviously Taylor did a solid amount of work and I thought it was definitely interesting enough to share.
Categories
Other
Previous Post Monday Bolts – 2.23.09
Next Post 25 Things about the Thunder
16 comments
  Livefyre
  • Get Livefyre
  • FAQ
Sign in
+ Follow
Post comment
 
Link
Newest | Oldest
Keith
Keith 5pts

@Taylor
I totally understand that what you're doing is already a great time investment. But thinking about it, I could probably work something up on my own using just one more piece of information from you.

1) You already are keeping track whether a "Could Have" assist is open, contested, and a three prong idea of where it went (key, jumper, three).

2) You also already have a measure of which passes led to points (recorded assists) with the same information.

All you would need is a quick little check mark to say who caught each pass. Since you are already marking down that a pass is being made, where it is being caught/shot, and whether it goes in, it doesn't seem much more work to mark down who got the shot.

If you don't wish to do the calculations yourself for this secondary part, I'd be willing to do it myself from your notes. It wouldn't be hard to figure out ast-shot% for each player once I know who is receiving passes and who is hitting shots. Then I could break it down by player to figure out who shoots the best and when (off of passes). From here, I could simply reference passes (with the information you already provide, I could know a great deal) with smart decisions, which is to say, reference whether passes are being made to high percentage situations, or whether passes are just being made.

Taylor
Taylor 5pts

@Keith
I like the idea as well of seeing who players are passing too. It would obviously be a more intelligent pass to kick it to Durant for three as compared to Watson. The only problem with that is that there are very few instances where the playmaker has two teammates open requiring him to make a more intelligent pass. In most cases simply making the pass to the open player is the smart play regardless. Recording who each pass is too and what there percentages are in the locations that they received the ball would be an awesome stat but I must admit that I am not up for the task! Haha.

DSMok1
DSMok1 5pts

Excellent work, Taylor! This is really interesting information, and can be useful in pseudo-quantifying those perpetually nebulous things that differentiate the decent player from the good or great player (besides the shooting stats).

Keith
Keith 5pts

The stat portion I was most intrigued by is the "Could Have Been Assist." It's great to finally get an idea of not just how well teams are shooting (a recorded assist), but also how much players are setting up teammates. A point I think one might need to address, however, is the percentage chance that specific passes lead to scores. For example, if Westbrook drives the lane and passes out to a waiting Weaver or Watson, chances are those passes will not end in points scored. However, if a pass is made to an open Durant or Green, the chance of points being scored is much higher.

Something to this effect would begin to quantify a player's decision making skills. Just passing to a player who shoots doesn't not make someone a playmaker, it just means that player got rid of the ball. By figuring out how many passes had a good chance of leading to points would say a lot more about how smart a player is playing.

Since you already have a great deal of statistics, it could probably be a function of those. Find shooting percentages of each player in specific situations regarding passes (getting a pass in the key, that leads to uncontested 2 and 3, or contested 2 and 3), then factor that into the passes thrown. Is a player getting assists because he is throwing the ball to the right teammate in a good position, or is he simply getting rid of the ball? This could significantly help determine a few things. One, how good is a player who dominates ball at helping teammates in actuality. Two, are players with low assist numbers selfish, or are they simply choosing quality over quantity with their passes. Three, is Westbrook a point guard or a shooting guard who is asked to pass a lot.

Taylor
Taylor 5pts

I do have a "High Post" and "Low Post" success rate on tap for tomorrow night's game. I will try that and then see if I can get more specific with it.

Nix
Nix 5pts

Awesome work Taylor...

I'd really like to see some sort of 'difficulty of post percentage'

Something that can tell us if a guy was being well guarded when he scored in the post.

John
John 5pts

@Taylor
If you have Tivo or DVR, you could even compile the stats more or less real-time, catching up to the game during commercial breaks & halftime. Cool stats, thanks for sharing this!

hoopinion
hoopinion 5pts

Royce :
I like that idea. In the end you could kind of come up with a basketball version of baseball’s OPS. It would be maybe the best way to gauge a player’s true offensive output. I may keep track of this myself for a few games and see how it turns out.

The main thing I've learned from trying to track this myself is just how many times television only shows you part of a possession. That and no matter how good I think my note taking is during the game, there's always something that doesn't reconcile with the official play-by-play the next morning.

Royce
Royce 5pts

hoopinion :
I have long advocated (and long failed to formulate a plausible plan to implement this change) replacing the assist column in the box score with one called “Points Created” which would be on the same scale as points scored thus including credit for the difference between assisting on two- and three-point field goals and would-be assists which result in made free throws rather than made field goals. Perhaps the added value isn’t worth the trouble.
Needless to say, I enjoyed this post very much.

I like that idea. In the end you could kind of come up with a basketball version of baseball's OPS. It would be maybe the best way to gauge a player's true offensive output. I may keep track of this myself for a few games and see how it turns out.

Royce
Royce 5pts

All in all Taylor, you did a fantastic job with this stuff and I think it's great. Clearly it's not all perfect, but it's a great way to think outside the box.

And let me see what I can do about that link.

Taylor
Taylor 5pts

Also, I want to do Per 36 minute averages when more data is compiled.

hoopinion
hoopinion 5pts

I have long advocated (and long failed to formulate a plausible plan to implement this change) replacing the assist column in the box score with one called "Points Created" which would be on the same scale as points scored thus including credit for the difference between assisting on two- and three-point field goals and would-be assists which result in made free throws rather than made field goals. Perhaps the added value isn't worth the trouble.

Needless to say, I enjoyed this post very much.

Taylor
Taylor 5pts

The stats are very subjective and I agree with all the comments. These categories make it very hard for their to be a definitive explanantion for when it should be rewarded.

The "Could Have Been Assist" does not mean that the player "should" have had an assist and it is simply an attempt to see who is making plays beyond the traditional assist stat. Obviously assist chances and actual assists will never be 100% or anything close to it. I am interested to see what the actual statistic would be: 50%? 65%? Who knows.

I am sure coaches keep many of these stats but we can't see what they are doing. It took me about two hours to complete and when comparing that to the fact that it would take me about 75-90 minutes to simply watch the game it didn't add too much time. Come on now Jax! Haha.

Against the Lakers I am going to keep track of Kobe, Lamar and Pau's stats as well just so we can have some more numbers to compare.

Jax Raging Bile Duct
Jax Raging Bile Duct 5pts

First, I think Taylor did a great job.
Second, I think Taylor needs more to do in life.

JK Taylor!

daniel
daniel 5pts

I can't read the link because Google is asking me to login.

Nevertheless, from your excerpts, sounds like a fascinating group of suggestions. Some of them lean towards being subjective in terms of how they would be awarded, so I'm not sure about that.

Coaches would probably like a few new stats. That would give them something to dangle in a player's face for incentive. Because once a number gets recorded and displayed, players will definitely play to the stat (like now when they are conscious about trying to get a double-double, for example).

Joe
Joe 5pts

Wow. This is sort of like the shorthand that Dean Oliver uses in his book to capture all of the stuff that happens in a game..only with some cool labels.

Trackbacks

  1. Detailed stats from the Thunder/Lakers game | Daily Thunder.com - Where Thunder Happens says:
    February 25, 2009 at 2:42 pm

    [...] know what each stat means, we’ll just go with the box score. And if you forgot what a CHBA is click here for a quick reference [...]

Back to Top

Headlines

  • VP Biden called Durant to thank him for donation
  • KD named All-NBA first team, Westbrook second team
  • Perk donates $25,000 to help build shelters in schools
  • Report: Mo Cheeks to interview with the Pistons
  • Thunder land the 12th pick in the 2013 draft
  • Thunder donate $1 million to aid with disaster relief
  • Kevin Durant donates $1 million to disaster relief
Daily Thunder
  • Home
  • About
  • Advertise
  • Contact

Copyright © 2008-2012 DailyThunder.com
Designed by iThemes Creative & Hosted by Site5