Half-Season Lineups' Stats – What worked for us, and what didn't

My thoughts on lineup stats:

In short: Like many things - they are basically very usefull and good, but only if used (and presented) properly.

I understand why you're against using lineup stats to draw conclusions about an individual player (like your TJ example). I totally agree that this is indeed a mis-use. You can't say any player "is balling" or "sucks" based on lineup stats!
I was trying to avoid that in my thread (I'll even try harder now ).

The valuable thing about lineup stats is they can tip you and give you a lead (if no more than that) on what works as a game plan, and more frequently - what doesn't...
For example, it can give you an idea about how the team performed with Twin-Tower lineups, Small-ball formations, lineups with more 3PT shooters, ISO-focused lineups, etc.

Variables of opposition and schedule are indeed very important and may bias the data considerably. I'd like to think that, given a large enough Sample Size (Which is NOT the case for most of our lineups I show here), this shall be evened-out statistically.

I can see your point about the emphasis on Starters and Closers, and you're mostly right there too. However, examining the rest of the lineups is not a waste of time IMHO - It's a good way to judge if a lineup stands out as producing great performance, and should be considered as a candidate for the more significant parts (like closing games). How else would you test lineups?

I did my best to provide sample-size... I waited half a season, didn't I?

So I'm going to keep on with this thread, anyway.

 

Agreed. It's a deal!

 

More stats - and now - an original one:

Pace

Pace is a very tricky stat, but an interesting one, nontheless.

The term may be misleading - It has nothing to do with how fast you run the court or move the ball. It measures only one single factor - the average length of possessions.
The shorter the possession - the more possessions you get to play in a game (i.e. the faster the Pace of the game is).

The reasons possessions are cut short vary widely and go both ways - A possession can be quickly lost in a TO, or it can start and end in a few seconds with a great steal & fast-break. We can run a short offensive play in purpose to try and get even when we are trailing, or the other team may foul us (even hack us... ) to stop the clock when we're ahead.

So if you think about it, it's obvious that a faster Pace in itself is neither "good" nor "bad". It all depends on how the players or lineups actually use their possessions... If their average possession yields a positive balance (i.e. they score more points then they allow in opponent's possessions), then a faster Pace is great - "the more (possessions), the merrier". If, on the other hand, each possession hurts our team, then of course we will benefit from a slower Pace...

The Pace is therefore quite literally a multiplier, a factor.

The reason I bring this stat here is because most of the stats posted in this thread were (/are/will be) calculated per 100 possessions, so for a specific lineup - having a high Pace factor would act as an amplifier for the lineup's other stats, whereas a low Pace would mitigate the effect of other stats.

OK then - Pace it is:



Findings:

1. Variance is not that big but lineup C, with Lin, Harden, Parsons, Casspi and Howard stands out as having a much higher Pace than others.

2. Strange enough - our closer lineup does NOT have a faster Pace as I anticipated...
Is it because our games do not tend to be close but either a blowout for us or against us?..

 

One more advanced stat (per 100 possessions) by lineups -
The Impact Estimate (PIE)

(I will not add any comments as yet - not for llaziness but because I must get some sleep before the game tonight)

PIE

 

I'm still gliding on that great win tonight (Blazers)... :grin:
BTW: Almost none of these frequent lineups was used tonight (we had to change things following the injuries).
Lineup C, that WAS used for some 5 minutes, was actually one of the few "bad" lineups of the night...

 

OK -

One last series of various stats for the 10 most frequent Rockets' lineups
as recorded at half-season mark (41 games).

I remind you that stats are all from this part of stats.nba.com.

Since there was not much of a discussion (an understatement...) about the previous stats,
I hereby bring all the rest of the measures at once, just to complete this data series.

Below you'll find the following stats' diagrams:

• True Shooting (definition & formula on top of the diagram)
• Rebound Percentage (explained in a sub-heading)
• Assist Ratio (explained in a sub-heading)
• Turnover Ratio (explained in a sub-heading)

Spoiler

Some Findings:

Spoiler

1. The not-that-frequently-used lineup H (8th in minutes played, but still only used for 33 minutes...), featuring AB, Garcia, Parsons, Casspi and Howard, leads by a large margin in the 3 main categories - Shooting % (TS% of 63.3), Rebounding (REB% of 65.5) and Assisting (AST Ratio of 22.0).
That would explain why this lineup ranked 1st in overall Offensive Rating.
--[Alert! This lineup's stats may be less significant due to a smaller sample size]

2. Runners-up in the 3 categories were:
In Shooting - Lineup A ("Start Bev") with 59.0% TS.
In Rebounding - Lineup C ("Lin & Casspi join the starters") with 57.2% (a surprising one! Both Twin Towers lineups are also close behind of course).
In Assisting - Lineup D ("Lin replaces injured Harden at the 2"), which is the expected suspect really - having 2 PGs.

3. Our least TO-prone lineup is by far team F, that also ranked #1 in Deffensive Rating as shown earlier - featuring Lin, Harden, Brewer, Jones and Howard, and used for a total of 55 minutes.
This lineup only turned the ball over 11.1 times per 100 possessions

4. Our most TO-prone lineup is, not surprisingly, one of the 2 Twin-Tower lineups - E, the one with Lin.

5. Of the 2 most used lineups A & B (the ones that actually DO provide a substantial sample size, and also embody our "Start Bev" vs "Start Lin" versions respectively) - The P-Bev alternative has a slight lead in all the 4 measures.

 

Hi again.

I looked at data from last 2 games (Spurs and Mavericks). Data was only relevant to one of the 10 leading lineups - The "Hardenless" starting lineup, marked in this thread as lineup D.

Interesting to note that in both of those games (wins, mind you!) that starting lineup was the one used the most, but was really not one of the most successful lineups. It's +/- contribution was actually negative in both games.

SAS game main lineups
(Sorted by number of minutes)

Spoiler

DAL game main lineups
(Sorted by number of minutes)

Spoiler

 

I figured I might as well go back to the 2 previous games -
the B2B losses to the Grizzlies:

In those 2 games it was our traditional starting lineup (A) that led in minutes.

The data is shown below.

MEM game 1 main lineups
(Sorted by number of minutes)

Spoiler

MEM game 2 main lineups
(Sorted by number of minutes)

Spoiler

(Spoiler - this lineup faired better, and was one of our better lineups in +/-)

 

Made the analysis for the Suns game too.

The data is shown below.

Suns game main lineups
(7 lineups that were used over 3 minutes, sorted by number of minutes played)

Spoiler

Notes:
★ At last!
Our starting lineups A (Bev) & B (Lin) performed superbly - better than most others - in pretty much everything:
Offensive and Defensive rating, PIE, shooting %, rebounding, assisting and TOs.

★ Howard-less lineups with 2 non-traditional bigs (D-Mo+Jones, Jones+Casspi)
were the worst lineups in this game.