21 replies to this topic

[thejohnnygold]

Since the topic of team turnovers has come up I got curious and compiled some data.  It's not the best comparison nor is it the best chart you've ever seen, but it does paint a pretty clear picture: these things matter.

 

 

What we've got here is a chart listing teams in order of last season's win totals with Defensive Efficiency and Team Assist to Turnover Ratios behind them.  I had to scale everything to get it on one chart because I am not good at charts.

 

What I see is a pretty clear correlation between winning and having good scores in these categories.  It is pretty easy to see that as Def. Eff. scores go up, wins go down.  A/To is a little less clear and context helps when looking at that aspect.

 

OKC and TOR clearly thumb their noses at this.  They are dual star-driven teams that each came up just short of the Finals.  Basically, outside of those 2 teams the only real contenders for the title all scored well in both of these categories.  I would also argue that OKC, TOR, and CLE all majorly ramped up their Def in the playoffs (moreso than other teams) which helped fuel their deep playoff runs.

 

(A lot of the teams with solid A/To ratios suffered major injury issues last season which, I think, partially explains the lackluster win totals for them.  Others are young, up-and-coming squads that I expect to make leaps soon.)

 

I'd love to hear everyone's thoughts on this and if any of our resident number crunchers wants to improve or add to this please, please do. 

 

I believe that Houston will drastically improve on the A/To aspect this year.  If they can also improve that Def Eff they will find themselves near the front of the line with the rest of the contenders.  Plenty of teams sport a decent A/To, but without the Def it does not guarantee wins.  The Def, on the other hand, is a pretty solid indicator of success by itself.

 

It is my belief that the NBA is moving in this direction and if we don't follow suit we will find ourselves left behind scratching our heads.

[Jatman20]

Nice graph. It's depicting the 15-16 season?

[Jatman20]

Yeah stats are crazy. GSW (4th at 101.1), Rockets (6th/7th at 100.2) and OKC (8th at 99.1) in Pace last season: ended up
Warriors 25th in TO's (14.9/gm), Rockets 26th (15.2/ gm), Thunder 27th/28th (15.5/ gm) all made the playoffs. I just hate
to use any Rockets numbers from last season due to the in-fighting of the two Alpha-Dogs.

[Jatman20]

I like the 3-point topic:Article: ["Do Three-Point Attempts Actually Lead To Winning In The NBA?"-Jesse Scudilla- Feb 17, 2016]


http://s9.postimg.or...Three_Graph.jpg

[Jatman20]

TPD Correl = 3-Point Differential Correlation
TP%D Correl = 3-Point % Differential Correlation
3PA Correl = 3-Points Attempted Correlation

Edited by Jatman20, 13 October 2016 - 02:40 AM.

[thejohnnygold]

Yeah stats are crazy. GSW (4th at 101.1), Rockets (6th/7th at 100.2) and OKC (8th at 99.1) in Pace last season: ended up
Warriors 25th in TO's (14.9/gm), Rockets 26th (15.2/ gm), Thunder 27th/28th (15.5/ gm) all made the playoffs. I just hate
to use any Rockets numbers from last season due to the in-fighting of the two Alpha-Dogs.

 

Yes, this is from last season and the stats are from John Hollinger at ESPN.

 

I used the ratio rather than volume stats so that pace would not matter.  Further, the ratio is important because many keep touting Harden's double digit assists while ignoring the mountain of turnovers he is accruing in the process.  I, for one, do not think they are mutually exclusive.

 

I agree that Houston's place in that chart is mucked up by all of last season's turmoil.  Yet, the point is still clear.  Those two categories along with point differential are huge indicators of playoff success.  One could argue that point differential is just a stat partially built off of the first two.

 

Every turnover has the opportunity to be a 5-6 point swing in the other direction.  (the loss of our opportunity to score plus the opponent getting that same opportunity).  Limiting those opportunities is a good idea--plain and simple.  This is the reason Morey went all in on producing turnovers a couple of years ago--they are game changers.  That experiment failed for a multitude of reasons.

 

Ultimately, if we can't produce on the defensive end all of this will be for nothing.  No amount of assists nor threes will overcome poor defense.

 

Regarding the article you referenced--after reading it I feel that there are two things to take away from it.  First, he repeatedly makes sure the reader understands it is 3 point shooting plus efficient defense that leads to winning.  Second, his final remark:

 

 

 

Overall, three-point shooting matters, but it's not the volume, necessarily, that leads to winning.

 

...it's how many of them go in.

 

To quote Kevin McHale from a few years back when asked about analytics and shooting efficiency, "well, it depends on who's shooting it".

 

This is why Kyle Wiltjer will be a valued NBA player.  He seems to already understand decent team defense which is a huge plus for him.  With some seasoning and a few cheeseburgers he could be a huge weapon in the growing 3-point war.

 

3 pt. shooting is great, but the variance can be a killer and as a coach, owner, GM, or whatever it is equivalent to gambling when you put all your eggs in that basket.  Even great shooters roller coaster from night to night.  5 out of 6 in one game and then 0-6 in the next with little rhyme nor reason to it.  The margin for error is that small.

 

Meanwhile, dunks and lay-ups are pretty consistent--unless playing teams with one of the few true rim protectors.  While dunks and lay-ups, etc. may lose to three point shooting on paper over the course of a season they definitely come out on top often and make for a very reliable offensive foundation.  Further, in a shortened season, aka the playoffs, that variance can make or break you.

 

(FYI, that same author wrote another article just a few months later titled, "The Miami Heat Need to Stop Shooting Three Pointers"  )

 

While I am all for 3 pt. shooting, I would rather build our team foundation on more solid principles.  Defense.  Rebounding.  Hustle.  Limiting turnovers.  Getting to the rim.  Taking open shots--wherever they are.  From there, incorporating quality three point shooters and getting them open looks is a bonus.

 

Take a look back at the A/TO list.  Look at the coaches in the top 10.

 

Popovich, Kerr, Rivers, Clifford, Stevens, Carlisle, Budenholzer, Lue, Wittman, and Skiles.  Overall, that is the cream of the crop of NBA coaches (OK, not Skiles and Washington fans would laugh at Wittman).  Is it any coincidence their teams lead the league in A/TO ratio?  I don't think so.

 

You will find those same names on the list of top 10 Def. Eff.

 

Popovich, Budenholzer, Vogel, Kerr, Stevens, Rivers, Spoelstra, Snyder, Clifford, Lue.  (7 out of 10 are the same and the three new ones who pop up aren't slouches)

 

Now, let's look at the teams with the most 3 pt. attempts.

 

Kerr, Bickerstaff, Lue, Clifford, Carlisle, Stotts, Budenholzer, Brown (PHI), Rivers, and Van Gundy.  (6 out of 10 the same)

 

We can see that more attempts do not equal more wins.  It depends on who's shooting...

 

Quite a few of these teams sport quality percentages and that is where the wins come in.  (Note that most of those teams also sport strong defensive mentalities)  Yet, Chicago, New Orleans, and Sacramento were all top 10 3pt% teams....not so much with the winning.

 

I guess the point I am belaboring here is that we can't just roll out a bunch of 3 pt marksmen, jack threes all game long and expect anything better than a one-and-done playoff showing.  If only it were that simple...

[Jatman20]

Yes, I think all the stats tend to dove tail into one another. The stereotypical teams that exploit fast breaking-uptempo
styles tend to be turn-over prone and poor defenders. Warriors have overcome the poor defenders label. How much
of it was Bogut and his slightly over 20 minutes/game. I didn't see much rim protection from the Warriors after he
went out.

[Jatman20]

Here is a link to an article that explains what may have been the biggest reason Houston dropped from 2nd seed to 8th seed.
I expect Bzdelik (with Morey's numbers crunchers to help him address the matter) to correct this area.
["One Team, One Stat: Rockets Regression"-John Schuhmann, Oct. 8, 2016]. I believe the author means
Defensive Rtg over Defensive Efficiency.

http://www.nba.com/a...3-point-defense

Edited by Jatman20, 13 October 2016 - 09:32 PM.

[Jatman20]

I don't want to take this thread hostage by talking 3's instead of TO's. I will just make this post and step aside.
Warriors regular season 15/16 3P%: Draymond (39%),Curry (45%),Klay (43%), Barnes (38%), Iggy (35%),
.........................................Barbosa (36%), Speights (39%)...7 rotational players with decent 3 pt shooters.

Rockets regular season 15/16 3P%: Harden (36%), Ariza (37%), Beverley (40%), JET (36%)...4 rot players.
It was no contest.
This year you take out Barbosa/Speights/Barnes and replace them with Durant (39%-15/16)

Rockets..........Player.............Last year...................Pre-Season........................................................................
..................Harden...................36%.............................39%..............................................................................
..................Ariza.......................37%.............................33%............(I expect % to go up).................................
.................Beverley..................40%.............................33%.............(I expect % to go up)................................
.................Dekker....................39%.....(Summer).........40%..............................................................................
.................Gordon...................38%..............................50%...............................................................................
.................Ryno.......................37%..............................48%...............................................................................

I'm not saying the Rockets are the Warriors; but I'm saying they can win about 56-60 games if they remain healthy.
Last year the Rockets played one Rook (Ariza), One Bishop (Bev), One Knight (D12) with a Queen (versatility of Harden).
This Year the air support is much greater: Gordon/Bev (2 Rooks), Ryno/Ariza (2 Bishops), Harden (Queen), Nene (Knight).
Teams play chase with the Warriors as a result of a "fear factor" (fear of their 3-pt shooting). As a result the Warriors
defense puts up good stats (as teams get out of their norm and jack up shots.....resulting in long defensive rebounds.
And stops. Honorable mention:McDaniels(40%), Prigioni (50%), Brewer (67%). Yes team numbers are likely to come down
in the regular season. Last year: Rush (41%), Looney (50%), McAdoo (50%), Ian Clark (36%). Rockets are good enough
to get a good playoff seeding (barring injuries). I won't promise playoff success. Not just yet.

Edited by Jatman20, 13 October 2016 - 10:24 PM.

[thejohnnygold]

Yeah, you could almost call it "bad luck" and not be too far from the truth.  (it is probably more correct to say we had "good luck" the year before when opponents kept missing wide open shots)

 

Then again, when you consider the subtle psychological aspects involved maybe it wasn't luck.  In '14-'15 we were hustling our butts off and our energy was usually pretty high.  D-Mo was wrecking people in the post (when he got the chance) and I think that can translate into making opposing teams (and shooters) maybe rush that little bit, or let their minds wonder (for that split second) if they can get the shot off...which will inevitably lower percentages.

 

On the flip side, it was no secret our team was a wreck.  Guys were phoning it in and opposing shooters likely had confidence not only that they could shoot relaxed, but that even if they did miss it wouldn't matter because Houston would roll over in the end and they were going to win.  Things like this will increase percentages.

 

There probably isn't much of any kind of actual data to support this, but I believe it to be true to one degree or another.

[Jatman20]

After 5 preseason games (some noticeable jet lag and introduction to full court pressure....vs Memphis)
Rockets are 6th best in TOV% at 13.4. I was watching the Boston-Knicks replay in which the commentator
claimed there are generally more TO's in preseason than in the regular season. Feels better to be
6th best than 6th worst, right now. Same can be said about the competition. Yes, it was the Sharks; but
when I said Sergio Llull might be more than adequate in the NBA....many scoffed at me and said he plays
in a sorry amateur league. Tell that to the Thunder who lost to that amateur team. I'm going to post on the
preseason impressions thread some of the team stats to this point. Mostly good.

Edited by Jatman20, 16 October 2016 - 11:38 PM.

[Sir Thursday]

@jg: Seems like what you actually want is a regression comparing Wins against various stats. For that you actually want to plot wins against whatever statistics you are trying to show are important and then do a linear regression. I'll have a go at making a few next time I get a free moment (don't hold your breath though - I'm pretty busy at the moment!)

 

ST

[thejohnnygold]

@jg: Seems like what you actually want is a regression comparing Wins against various stats. For that you actually want to plot wins against whatever statistics you are trying to show are important and then do a linear regression. I'll have a go at making a few next time I get a free moment (don't hold your breath though - I'm pretty busy at the moment!)

 

ST

 

That would be great!  I know my chart is rudimentary at best--both in the data being used and how it is organized.

 

Not holding my breath, but I do hope you find the time 

[Sir Thursday]

OK, had a quick go, grabbed the stats off basketball-reference. If you look at the various tabs, you'll see a few different stats plotted against how many games each team won: https://public.table...splay_count=yes

 

I did ORB%, DRB%, TOV% and Opp TOV% to start with. Main findings:

 

• DRB% - increasing by 1% seems to be worth about 2 wins.
• ORB% - increasing by 1% seems to be worth about 0.5 wins.
• TOV% - decreasing by 1% seems to be worth about 5.6 wins.
• Opp TOV% - there's basically no correlation with winning.

Now, before you run with these, I would be irresponsible if I didn't make a few caveats:

• I've assumed a linear regression, which is almost certainly not the case for something as complicated as basketball. So the actual effects are likely to be different than described.
• None of the correlations are incredibly strong - TOV% is the only one that would pass most statistical significance tests.
• This sort of analysis is inherently susceptible to a phenomenon known as p-hacking: because of the number of regressions we're doing, at some point you would expect a correlation to appear in the data just by chance. You're supposed to do a Bonferoni correction to adjust for that - I haven't bothered here because this is very informal. But in order to draw more rigorous conclusions we should probably do that at some point.

Anyway, have fun looking at the data, and let me know if you want any more!

 

ST

[thejohnnygold]

OK, had a quick go, grabbed the stats off basketball-reference. If you look at the various tabs, you'll see a few different stats plotted against how many games each team won: https://public.table...splay_count=yes

 

I did ORB%, DRB%, TOV% and Opp TOV% to start with. Main findings:

 

• DRB% - increasing by 1% seems to be worth about 2 wins.
• ORB% - increasing by 1% seems to be worth about 0.5 wins.
• TOV% - decreasing by 1% seems to be worth about 5.6 wins.
• Opp TOV% - there's basically no correlation with winning.

Now, before you run with these, I would be irresponsible if I didn't make a few caveats:

• I've assumed a linear regression, which is almost certainly not the case for something as complicated as basketball. So the actual effects are likely to be different than described.
• None of the correlations are incredibly strong - TOV% is the only one that would pass most statistical significance tests.
• This sort of analysis is inherently susceptible to a phenomenon known as p-hacking: because of the number of regressions we're doing, at some point you would expect a correlation to appear in the data just by chance. You're supposed to do a Bonferoni correction to adjust for that - I haven't bothered here because this is very informal. But in order to draw more rigorous conclusions we should probably do that at some point.

Anyway, have fun looking at the data, and let me know if you want any more!

 

ST

 

That's awesome--Thanks!

 

Since you are offering, I wouldn't mind seeing Assist/turnover ratio and scoring margin done this way.

 

The ORB% is a weird one.  One would think that a team with poor shooters would be more inclined to crash the offensive glass because they know there is a good chance for a rebound.  I find it fascinating that Philly and New Orleans are so bad in this category.  Looking at their players I would think they would be better.  Shows what I know....

[Jatman20]

True, all stats as a whole show whole team performance (or lack there of, in the form of wins/loses).

TOV%: OKC 14.0 (55 wins); Houston 14.2 (41 wins), Lakers 12.5 (17 wins).

[Jatman20]

In post 4 of this thread I linked a Scudilla article:

Team(15/16).........3P%..(rank)..................Opp3P%..(rank).................3PA...(rank).............Opp3PA...(rank).....................
GSW................... .416..(1st)..................... .332..(2nd)........................ 31.6..(1st)................... 23.5..(13th)...........................
Clev..................... .362..(7th).................... .347..(13th)....................... 29.6..(3rd)................... 22.7..(7th).............................
Spurs.................. .375..(2nd).................... .331..(1st)......................... 18.5..(26th)................. 19.9..(1st)............................
Per basketball-reference.com

I expect the Rockets to: Excellent 3P% (about 38%+), Excellent 3PA (about 30-40 attempts/game), and if Bzdelik does
his job a greater defense of opponents 3-pointers......especially the corner three. In 14/15 the Rockets placed emphasis
on guarding the corner 3's. In 13/14 the Rockets were one of the worst teams in protecting the corner three's.

Edited by Jatman20, 19 October 2016 - 01:36 AM.

[Jatman20]

In post #16....note coach Byron Scott (Lakers) was not particularly fond of team 3-point shooting. Thus 17
wins may have been an end result.

Byron Scott's philosophy on teams shooting 3-pointers: (I will get off my 3-point shooting soap box now)

http://www.cbssports...s-championships

Edited by Jatman20, 19 October 2016 - 12:02 PM.

[thejohnnygold]

Conveniently, Henry Abbott (Truehoop) just posted this article about elite defense being a better indicator of championship success than elite offense.  This article needs to be put on the wall in the Rockets' locker room.

 

I don't care how many threes we shoot, or make, this season.  If the guys don't get after it on defense then they are just pretending to want to win a championship and, as fans, we will spend this year trying to talk ourselves into something that simply isn't true.

 

Sorry to bring so much doom and gloom.  It's just so depressing to me that this is an issue at all and I fear that this team is set up to fail; albeit, in a glorious-guns-blazing-sort-of-way.

 

Zach Lowe agrees in his most recent article (which pointed me to the article linked above) :

 

 

The only question that matters -- and one that got a little more pressing with the news from our Calvin Watkins that Patrick Beverley may need knee surgery: Just how bad is Houston's defense going to be?

 

There is a lot of evidence that an elite defense is a slightly more powerful predictor of championship contention than an elite offense, and the same may be true on the downside. Only seven teams that fell into the bottom five in points allowed per possession made the playoffs over the past 20 seasons, according to data compiled by ESPN Stats & Information. Those teams ranked about third on average in offensive efficiency.

The prognosis gets better if you are merely not terrible; a full 32 teams, about 1.5 per season season, squeaked into the playoffs with bottom-10 defenses. Meanwhile, a bunch more -- 42 -- got in despite bottom-10 offenses over that same 20-year stretch.

Houston is going to score the hell out of the ball. If they stay healthy and trudge closer to 20th in overall defense, they should win enough games -- something in the high-40s -- to secure a spot. That's not easy for any team featuring Ryan Anderson, an unhidable saboteur, and it definitely won't be easy if Houston gets the comatose version of James Harden and Eric Gordon. Some of these guys have scary injury histories.

But Harden is in shape, and the Rockets have enough solid defenders -- including two centers -- to achieve some minimum level of competence. Having perhaps the second-best offense in the league will help; it's easier to set your defense and get stops after a bucket -- or even better, a free throw.

 

Yes, the offensive improvement will help, but if we aren't going all out on defense than we are just going through the motions and not contenders for anything other than a modest playoff bonus check before bowing out and letting the teams who want to be there play.

[Sir Thursday]

Conveniently, Henry Abbott (Truehoop) just posted this article about elite defense being a better indicator of championship success than elite offense.  This article needs to be put on the wall in the Rockets' locker room.

 

I don't care how many threes we shoot, or make, this season.  If the guys don't get after it on defense then they are just pretending to want to win a championship and, as fans, we will spend this year trying to talk ourselves into something that simply isn't true.

 

Sorry to bring so much doom and gloom.  It's just so depressing to me that this is an issue at all and I fear that this team is set up to fail; albeit, in a glorious-guns-blazing-sort-of-way.

 

Zach Lowe agrees in his most recent article (which pointed me to the article linked above) :

 

 

Yes, the offensive improvement will help, but if we aren't going all out on defense than we are just going through the motions and not contenders for anything other than a modest playoff bonus check before bowing out and letting the teams who want to be there play.

 

"Just posted" might be a bit of a stretch, seeing as how that article was from 2010...

 

ST