First of all, Happy New Year, everyone! And I won't blame you if you don't want to look at these numbers after last night's disappointing (to say the least) performance by the Monta-less Warriors (this is the part where I say, hey, small sample size, just one game, yada yada, but still...). But for those who do venture beyond the jump, I do have something new for you...
So, here's the new thing:
The math behind this is actually relatively simple. I take a prior average for the % each play is run and the efficiency (PPP) for each team. Right now, I'm using last season's data, but as the season progresses, I will start to blend the current season. The expected rate and PPP is then the simple average between the offense of one team and the defense of the other team for each play type. Here is what the expected Synergy stats looked like for last night's game:
|
|
EXPECTED |
|||
|
|
GSW |
PHI |
||
|
PLAY |
% |
PPP |
% |
PPP |
|
ALL |
|
0.93 |
|
0.95 |
|
BALL |
13.0% |
0.87 |
10.3% |
0.81 |
|
CUT |
7.1% |
1.18 |
8.1% |
1.21 |
|
HAND |
2.3% |
0.95 |
2.7% |
0.91 |
|
ISO |
12.8% |
0.85 |
12.3% |
0.85 |
|
OTHER |
6.5% |
0.41 |
6.9% |
0.45 |
|
POST |
8.5% |
0.82 |
9.4% |
0.92 |
|
REB |
6.1% |
0.99 |
6.2% |
1.06 |
|
ROLL |
5.2% |
0.99 |
4.6% |
0.97 |
|
SCREEN |
5.0% |
0.89 |
4.4% |
0.89 |
|
SPOT |
19.4% |
1.01 |
19.3% |
0.98 |
|
TRANS |
12.2% |
1.16 |
14.1% |
1.18 |
The top line (ALL) gives the overall predicted team efficiency (0.93 for GSW and 0.95 for PHI). Now let's look at what really happened:
|
PHI@GSW |
GSW |
PHI |
||||
|
|
% |
# |
PPP |
% |
# |
PPP |
|
ALL |
100.0% |
103 |
0.75 |
100.0% |
106 |
0.95 |
|
BALL |
5.8% |
6 |
1.17 |
4.7% |
5 |
0.60 |
|
CUT |
11.7% |
12 |
0.58 |
5.7% |
6 |
1.17 |
|
HAND |
N/A |
N/A |
N/A |
2.8% |
3 |
1.00 |
|
ISO |
19.4% |
20 |
0.50 |
18.9% |
20 |
0.70 |
|
OTHER |
1.0% |
1 |
0.00 |
1.9% |
2 |
1.00 |
|
POST |
8.7% |
9 |
0.78 |
8.5% |
9 |
1.00 |
|
REB |
8.7% |
9 |
0.89 |
6.6% |
7 |
0.86 |
|
ROLL |
4.9% |
5 |
0.80 |
2.8% |
3 |
1.33 |
|
SCREEN |
6.8% |
7 |
0.71 |
1.9% |
2 |
1.50 |
|
SPOT |
17.5% |
18 |
0.89 |
26.4% |
28 |
0.96 |
|
TRANS |
15.5% |
16 |
0.81 |
19.8% |
21 |
1.10 |
As you can see, PHI lived up to their end of the bargain, going for a 0.95 PPP exactly. But we, um, did not quite live up to expectations, managing a pathetic 0.75 PPP. Much of this was due to going 0.5 PPP on ISO plays and an extremely low 0.58 PPP on cut plays (which are usually among the most efficient plays a team can get). Also, our efficiency in transition was only 0.81 PPP compared to the expected 1.16. In short, we played like crap.
Here's a Treemap representation of the game. The area of each tile represents the %-rate and the color represents the efficiency (red is hotter, blue is colder). Click on each tile to pop-up the % and PPP.
These are the reference rates on offense and defense from last season used to calculated the expected numbers shown above.
|
|
GSW |
PHI |
||
|
|
% |
PPP |
% |
PPP |
|
ALL |
100.0% |
0.95 |
100.0% |
0.94 |
|
BALL |
13.7% |
0.88 |
11.5% |
0.83 |
|
CUT |
7.4% |
1.19 |
7.6% |
1.21 |
|
HAND |
2.2% |
0.99 |
3.2% |
0.80 |
|
ISO |
14.0% |
0.85 |
13.9% |
0.82 |
|
OTHER |
6.6% |
0.39 |
6.1% |
0.50 |
|
POST |
6.6% |
0.76 |
8.8% |
0.93 |
|
REB |
6.1% |
0.98 |
5.9% |
1.05 |
|
ROLL |
5.2% |
0.96 |
5.2% |
0.91 |
|
SCREEN |
4.6% |
0.86 |
4.0% |
0.83 |
|
SPOT |
20.2% |
1.05 |
19.5% |
0.94 |
|
TRANS |
13.4% |
1.20 |
14.3% |
1.19 |
|
|
GSW |
PHI |
||
|
|
% |
PPP |
% |
PPP |
|
ALL |
100.0% |
0.92 |
100.0% |
0.88 |
|
BALL |
9.0% |
0.78 |
12.3% |
0.86 |
|
CUT |
8.5% |
1.21 |
6.7% |
1.17 |
|
HAND |
2.1% |
1.01 |
2.4% |
0.90 |
|
ISO |
10.6% |
0.88 |
11.5% |
0.85 |
|
OTHER |
7.6% |
0.40 |
6.4% |
0.42 |
|
POST |
9.9% |
0.91 |
10.3% |
0.87 |
|
REB |
6.4% |
1.07 |
6.0% |
0.99 |
|
ROLL |
3.9% |
1.02 |
5.2% |
1.01 |
|
SCREEN |
4.7% |
0.94 |
5.4% |
0.91 |
|
SPOT |
19.1% |
1.02 |
18.6% |
0.97 |
|
TRANS |
13.8% |
1.17 |
10.9% |
1.12
|
Here are the league average efficiencies from last season. You can use these to compare the values in the tables above. Blue line represents GSW's average.
Here are the league average rates (% of time each play was run) from last season. GSW is in blue.
0 recs | 13 comments
As always
Appreciate the analysis and work you put in to this.
BeePill - January 1, 2012
Appreciate the analysis and work you put in to this.
Yeah, good job.
Is it worth doing the early season games using last years data though? Seems like most teams change enough to make the results flawed? About halfway thru the season enough new data will have been collected to make the results morerelevantt to the current situation?
Skeptic con Urquell - January 1, 2012
I think it’s worth doing, yes. Some teams change more than others, but most teams are fairly similar. Philly is. Like I said, as more data comes in, last season will get less weight.
Evanz - January 1, 2012
Evanz
first -i want to thank you for provided stats
second -may i ask you put some numbers about W’s PnR O -for -let’s say last 86 games
with all this PPP % -how may times big man is finisher -who of current ballhandlers are better at this etc.
Lat We N Trash - January 1, 2012
It’s actually there. BALL is when the ball handler (Curry or Ellis usually) on a PNR shoots the ball, and ROLL is when the rolling man (usually Lee) shoots the ball.
I summed up a lot of this after last season:
http://www.goldenstateofmind.com/2011/5/28/2194653/analyzing-warriors-offense-in-2010-11-a-synergy-perspective
Evanz - January 1, 2012
Updated the post with a Synergy Treemap
Evanz - January 1, 2012
The way I see it so far, the Warriors could make it as an 8 seed.
But I think Portland, Denver, Memphis, L.A.C., and Maybe Houston are still better. I haven’t seen Houston play this year yet though.
kenntoe - January 1, 2012
Lowry is real good
Evanz - January 1, 2012
So, not lookin' good.
kenntoe - January 1, 2012
Memphis just lost ZBo
Duby Dub Dubs - January 4, 2012
Great stuff
The analysis, not the subject.
These Expected Synergy Stats are great. That way we can see where we played well, where we sucked, and can incorporate observation/scouting to find out why that was the case. Great stuff, looking forward to these after every game! No pressure or anything.
lincypoo i wuv u - January 1, 2012
"Much of this was due to going 0.5 PPP on ISO plays..."
I think this is definitely the most significant point about this game.
In both the Knicks (without Curry) and Sixers (without Ellis) games, the Warriors used more ISO plays than expected. Against the Knicks, they were at .8 ppp; against the Sixers, .5 ppp.
But one other thing that really stands out in the numbers is the number of players relying on others to create shots for them (via Hoopdata – (vs. Knicks; vs. Sixers): against the Knicks, only one player (Biedrins, 1-1 at the rim of course) got 100% of their scoring output on assists; against the Sixers five players (Rush, Thompson, McGuire, Brown, Biedrins) had 100% of their scoring output on assists. The way this offense has worked – "freestyle" – you simply can’t have two perimeter players (58 minutes combined) relying entirely on others to create points while the point guard’s usage rate is at 33.9%.
On top of that, you had Ekpe Udoh with a usage rate of 31.8% and going 2-for-8 away from the rim.
One thing Ellis does is create more shots for himself and others at the rim, which is important. And Ellis draws attention from the defense in ways that would’ve been helpful against the Sixers. But having so few players creating points for themselves isn’t ever going to work for a team whose offense is so perimeter-oriented to begin with.
Part of this is definitely that the team needs to actually work on having an offense that gets players open for shots, especially for nights that Ellis is out/off… but part of it is also that the Sixers presented the Warriors with outstanding perimeter defenders on a night while one of the Warriors’ top creators was out. It’s a flaw in the roster that I’m sure other teams will pick up on, but it’s also such a unique circumstance for this roster that I’m not sure we can read too much into it.
OK, I need to just get over that game now and move on. ;)
Nate Parham - January 1, 2012
EvanZ passed this on via Twitter and it's worth a read re: Udoh
http://www.countthebasket.com/blog/2008/03/06/diminishing-returns-for-scoring-usage-vs-efficiency/
It’s hard to expect Udoh to score efficiently at that high a usage rate, especially when he’s shooting that many shots away from the rim.
Nate Parham - January 1, 2012
You must Login with your SB Nation account and be a member of Golden State Of Mind to post a comment.