Gordon Hayward: Pull Up/Drive Weekly Update

[NUMBERICA]

Doesn't the info in link 2 justify Sloan's overly aggressive help defense that left the three open more than usual? It's funny how the more analytics catches up to Sabermetrics the more they're finding out that coaches already knew most this stuff.

I think you have to be case-specific with things like this. In Sloan's case, the scheme seemed to be to indiscriminately help off the corners and perimeter which works great when the shot doesn't go in. When it does go in, it doesn't seem that the Jazz could adapt.

 

[NUMBERICA]

https://stats.nba.com/playerTrackingPassing.html?pageNo=1&rowsPerPage=25

Hockey assist is listed as "Secondary Assists per game".

Kobe Assist (Kirk Goldsberry of courtvisionanalytics.com) would have to be included in the definition of when a drive ends. I haven't found anyone releasing their data and doubt it's included in GVC's #'s even though it's a very important part of offense building. We'd probably have to count Kobe Assists and team offensive rebounding rate the long way and add them in somehow.

This stat might mean something more to me when we get a larger sample size as I'm not impressed with a number of people that outperform Hayward by this metric (Trevor Ariza #1)? The hard thing about so-called hockey-assists is that the person that is credited with one is particularly at the mercy of consequence once the ball leaves their hands. In other words, their credit is more contingent on their surrounding circumstances than on the circumstances that they create.

Still, it's interesting.

 

[franklin]

This stat might mean something more to me when we get a larger sample size as I'm not impressed with a number of people that outperform Hayward by this metric (Trevor Ariza #1)? The hard thing about so-called hockey-assists is that the person that is credited with one is particularly at the mercy of consequence once the ball leaves their hands. In other words, their credit is more contingent on their surrounding circumstances than on the circumstances that they create.

Still, it's interesting.

Maybe it depends on what play, but there are several hockey assist by design swing passes that should credit the driving player for imposing his game upon a defense, forcing it to help and then help the helper. The Spurs run these exceptionally well, and Danny Green is racking up assists off the back of Tony Parker's work. All he has to do is roam between spots behind the arc and make the easy pass if given or take the easy shot. He's not creating for others though, TP is.

 

[GVC]

Maybe it depends on what play, but there are several hockey assist by design swing passes that should credit the driving player for imposing his game upon a defense, forcing it to help and then help the helper. The Spurs run these exceptionally well, and Danny Green is racking up assists off the back of Tony Parker's work. All he has to do is roam between spots behind the arc and make the easy pass if given or take the easy shot. He's not creating for others though, TP is.

Ideally, you want to know outcomes by play initiation. That is, how well does a team do when they initiate their offense through an iso/PnR/Post-up (or, in a perfect world, what is the expected team outcome of an iso/PnR/Post-up of player x when surrounded by league average OR replacement level players). The data being collected now could and probably is being used to measure this. If the league is already making so much data available in the first season with the cameras, I think it's safe to assume we'll have much better data/stats in coming seasons (or even later this season).

The available data still provides us with some of the context missing in the current box score. Despite the limitations of the stats- I think it's safe to assume drives only includes points scored and assists credited to the player driving, and spot-ups doesn't include fouls drawn...neither takes into account the shot clock situation when the shots are taken, and leverage matters- they give us some useful information.

 

[GVC]

The BIG project the data might make possible, depending on the sophistication of network estimation techniques (of which I am incredibly ignorant), is the following:

Define a player based on a set of scores over a vector of key attributes/abilities (handle, halfcourt creation, transition defense, rim protection, on-ball defense, etc., etc.). Ideally, a reasonable number of attributes/abilities would account for the variation in player value (say 6-12). After defining individual players, test how well various 5-man combinations (defined by the joint distribution of the 5 players' attributes) perform against various 5-man combinations.

The goal here is to measure a player's context-specific value (the previous post was trying to get at a player's context-neutral value). In so doing, front offices, fans, etc. can get a better idea of what players would best complement the current talent on the roster and help to make cost/benefit decisions, given the available free agents (own team and other team), trade opportunities and the makeup of opposing teams.

 

[franklin]

It's nice that they're adding spacing elements into some analysis as well to handicap good play on bad teams and vice versa.

 

[GVC]

It's nice that they're adding spacing elements into some analysis as well to handicap good play on bad teams and vice versa.

I noticed you referencing some of Kirk Goldsberry's shooting stats. He almost certainly has the data to add a couple dimensions, space between shooter and nearest defender and off-dribble/spot-up, to his analysis.

 

[Jazz Spazz]

GVC, I enjoyed some of the stuff in the second link you provided in the first post.

Particularly the examples of plays that are normally run one way, and the defense gets used to it, but then they start the play that way then throw a wrinkle in and do something different in step 3 or 4 of the play progression and it confuses the defense. I like stuff like that and find it interesting. I know basketball and any other sport has a physical element, but if you can master the mental part you can succeed. It reminded me of football and disguised coverages by the defense, or offensive plays that are set up the same but you can do 4 or 5 things with the same set.

/ramble

 

[GVC]

Gordo's been an absolute beast on pull ups through the first 2 games this week. 12-15 from 2 and 1-2 from 3 (per NBA.com's player tracking stats).

 

[[size/HUGE] fixed [/size]]

He needs to shoot somewhere in the neighborhood of 6 threes a game. Minimum.

Godspeed Trey Burke.

 

[jope]

So, if gordon can play like this all season, we are giving him a significant raise. His comparable players in the pts/reb/asts are both high salary guys. Good to see from Gordon, now lets just see if he can sustain this for an entire season.

 

[Litany]

This is a great thread.