Golden State Warriors Four Factors Summary

I've been doing a lot of work at my blog recently on determining "adjusted" forms of a variety of different metrics. Most recently, I introduced a new metric that I call Adjusted Four Factor +/- or A4PM, for short. The idea here is to go underneath the hood of adjusted +/- and see how each player creates value for his team via the four factors (on offense and defense). Rather than regurgitate all the gory technical details here, I recommend you go to my blog for that, or just click after the jump to see a summary of results specific to the Warriors (which I did not present over there).

Briefly, the four factors (FF) are eFG%, TOV%, OREB%, and FTA/FGA. I've split the factors into offensive and defensive components. To give an example of how to interpret a rating, suppose that a player has a +1.0 rating in offensive eFG% (OEFG). This means that while that player is on the floor, we can expect a roughly 1-%-point increase in team-level eFG% on offense relative to an average team. Through a regression of the FF to RAPM, I also determined how each factor contributes to overall player value. This is given by the following equation:

A4PM = 1.77*eFG(own) - 1.60*eFG(opp) + 0.14*FTA/FGA(own) 

-0.11*FTA/FGA(opp) - 1.37*TOR(own) + 0.80*TOR(opp) 

+ 0.31*ORR(own) - 0.41*ORR(opp)

So, going back to the player with a +1.0 rating on eFG%(own), we can see that such a rating would contribute +1.77 to his A4PM rating. The correlation between A4PM and RAPM is very high (R^2=0.88), so we can roughly say that this player would have about the same amount of adjusted +/- value just based on his adjusted eFG% rating alone. Clearly, some factors are more important than others. In particular, eFG is by far the dominant factor, followed by TOV%, and then ORR and FTA/FGA somewhat further behind.

With that brief introduction let's now take a look at how the Warriors stack up in each factor. These ratings were done on the past 2 seasons plus games through about 1 week prior to the ASG. You can see the total possessions accounted for in the first table below. To get Klay Thompson in here, I made the minimum threshold 900 possessions. There were 440 players above this threshold, and each factor is listed along with a percentile rank (%-RK) based on this number.

The last column in each table gives the component of A4PM for that factor. In other words, the factor rating multiplied by the coefficient determined by the regression. This is the rating that really matters. The +/- rating for a 5-man unit can be predicted by adding up these 5*8=40 individual ratings (O1...D4).

Offense

eFG%

Curry, Udoh, and Lee make major contributions here. Monta is positive. Rush and Nate appear to be playing well this season, but according to this, we should consider it either an anomaly or major surprise. Dominic McGuire is a zero on offense.

%-RK	NAME	POS	POSS	OEFG	O1
96.4%	Stephen Curry	1.1	12333	0.87	1.50
90.7%	Ekpe Udoh	4.8	3040	0.53	0.92
88.0%	David Lee	4.2	13056	0.47	0.81
72.7%	Monta Ellis	1.8	13931	0.18	0.32
56.8%	Klay Thompson	3.0	913	-0.02	-0.04
48.0%	Dorell Wright	3.2	10533	-0.15	-0.26
37.5%	Andris Biedrins	5.0	5157	-0.26	-0.46
37.0%	Kwame Brown	5.0	4737	-0.27	-0.46
31.4%	Nate Robinson	1.0	4924	-0.35	-0.60
28.2%	Brandon Rush	2.5	9724	-0.42	-0.73
10.0%	Dominic McGuire	3.9	2535	-0.92	-1.60

TOR%

Udoh looks pretty good here, too. Note that the difference in (adjusted) turnover rates between Udoh and Biedrins equates to a whopping 1.6 points per 100 possessions.

%-RK	NAME	POS	OTOR	O2
90.2%	Ekpe Udoh	4.8	-0.52	0.72
83.6%	Brandon Rush	2.5	-0.43	0.59
69.5%	Nate Robinson	1.0	-0.20	0.28
67.7%	David Lee	4.2	-0.16	0.23
66.8%	Dorell Wright	3.2	-0.16	0.22
66.6%	Stephen Curry	1.1	-0.16	0.22
46.1%	Kwame Brown	5.0	0.07	-0.10
37.5%	Monta Ellis	1.8	0.20	-0.28
35.0%	Klay Thompson	3.0	0.23	-0.32
22.0%	Andris Biedrins	5.0	0.47	-0.65
8.0%	Dominic McGuire	3.9	0.99	-1.37

OREB%

Dominic looks good here, but compare the positive value of his offensive rebounding (0.3) to the negative value of his eFG% (-1.6) and TOV% (-1.37), and hopefully, you better appreciate the problem. And you surely know about our rebounding woes, but you probably assumed David Lee had a better overall impact on offensive rebounding than Ekpe Udoh. Perhaps, not, according to these results.

%-RK	NAME	POS	ORR	O3
81.8%	Dominic McGuire	3.9	0.84	0.30
63.0%	Dorell Wright	3.2	0.30	0.11
59.5%	Ekpe Udoh	4.8	0.21	0.07
49.1%	Kwame Brown	5.0	-0.05	-0.02
44.8%	Andris Biedrins	5.0	-0.17	-0.06
36.4%	Monta Ellis	1.8	-0.35	-0.12
30.2%	Nate Robinson	1.0	-0.52	-0.18
19.3%	David Lee	4.2	-0.86	-0.30
8.4%	Stephen Curry	1.1	-1.34	-0.47
5.2%	Brandon Rush	2.5	-1.63	-0.57
2.7%	Klay Thompson	3.0	-1.91	-0.67

FTA/FGA

We all know this team has issues getting to the line. Only Monta is above league average. Fortunately, this factor is not contributing as much as the others.

%-RK	NAME	POS	OFTR	O4
63.2%	Monta Ellis	1.8	0.20	0.03
51.1%	Dominic McGuire	3.9	-0.07	-0.01
33.9%	Brandon Rush	2.5	-0.54	-0.07
29.3%	Kwame Brown	5.0	-0.63	-0.08
26.8%	Stephen Curry	1.1	-0.69	-0.09
26.4%	Klay Thompson	3.0	-0.70	-0.09
17.5%	David Lee	4.2	-1.06	-0.14
10.9%	Nate Robinson	1.0	-1.53	-0.20
8.6%	Dorell Wright	3.2	-1.69	-0.22
8.4%	Ekpe Udoh	4.8	-1.83	-0.24
3.6%	Andris Biedrins	5.0	-2.46	-0.32

Defense

eFG%

Probably not a surprise seeing Udoh and McGuire at the top of this list (here's where Dom offsets some of his offensive liabilities). I am surprised to see Klay so high, but this could definitely be a sample size issue. The fact that 4 of our 5 starters going into the season were so far below simply treading water is a bad, bad sign. When you hear me constantly talking about defense being important, well, here you go.

%-RK	NAME	POS	DEFG	D1
98.2%	Ekpe Udoh	4.8	-1.24	1.99
95.5%	Dominic McGuire	3.9	-0.96	1.54
94.8%	Klay Thompson	3.0	-0.92	1.47
61.1%	Dorell Wright	3.2	-0.12	0.19
53.6%	Nate Robinson	1.0	-0.04	0.06
44.5%	Brandon Rush	2.5	0.06	-0.10
26.8%	Stephen Curry	1.1	0.30	-0.48
18.9%	Andris Biedrins	5.0	0.47	-0.76
13.0%	David Lee	4.2	0.61	-0.98
12.0%	Kwame Brown	5.0	0.64	-1.02
9.3%	Monta Ellis	1.8	0.70	-1.12

TOV%

Here again you see Klay at the top, but this is almost certainly a sample size issue. Monta's (and GSW's) gambling style of defense certainly leads to more opponent turnovers (although maybe not as many as you'd predict, for example, compare his 0.47 rating to Tony Allen's 1.90), but does not offset the highly negative defensive eFG ratings.

%-RK	NAME	POS	DTOR	D2
99.1%	Klay Thompson	3.0	1.79	1.48
88.6%	Ekpe Udoh	4.8	0.62	0.52
83.6%	Monta Ellis	1.8	0.47	0.39
80.2%	Dorell Wright	3.2	0.42	0.35
79.1%	Nate Robinson	1.0	0.39	0.33
74.3%	Stephen Curry	1.1	0.31	0.26
52.3%	Andris Biedrins	5.0	0.03	0.03
35.0%	Kwame Brown	5.0	-0.18	-0.15
22.3%	Brandon Rush	2.5	-0.36	-0.30
12.0%	David Lee	4.2	-0.51	-0.43
10.9%	Dominic McGuire	3.9	-0.54	-0.45

OREB%

This is a frightening table, if you're a Warriors fan. Once again, Udoh is surprisingly positive even though it doesn't show up in the box score. This also highlights the point that many of us make about David Lee's rebounding numbers being mostly hollow (he ranks in the 26th %-ile). A backcourt of Monta and Steph makes positive team rebounding difficult to achieve. But hey, throw in David Lee and...(that's really our "core"?)

%-RK	NAME	POS	DRR	D3
93.9%	Kwame Brown	5.0	-1.12	0.49
91.6%	Andris Biedrins	5.0	-1.01	0.45
79.1%	Ekpe Udoh	4.8	-0.54	0.24
54.5%	Dominic McGuire	3.9	-0.04	0.02
25.9%	David Lee	4.2	0.57	-0.25
22.5%	Brandon Rush	2.5	0.69	-0.30
19.5%	Dorell Wright	3.2	0.76	-0.34
11.8%	Stephen Curry	1.1	1.04	-0.46
8.2%	Nate Robinson	1.0	1.19	-0.52
6.1%	Monta Ellis	1.8	1.30	-0.57
0.9%	Klay Thompson	3.0	2.54	-1.12

FTA/FGA

Aside from offensive eFG, this is Lee's best category. Unfortunately, it's worth very little.

%-RK	NAME	POS	DFTR	D4
76.1%	David Lee	4.2	-0.60	0.07
75.7%	Kwame Brown	5.0	-0.59	0.07
49.3%	Andris Biedrins	5.0	0.12	-0.01
47.3%	Ekpe Udoh	4.8	0.17	-0.02
40.5%	Dorell Wright	3.2	0.31	-0.03
35.5%	Stephen Curry	1.1	0.44	-0.05
27.5%	Nate Robinson	1.0	0.79	-0.09
24.8%	Brandon Rush	2.5	0.88	-0.10
20.5%	Monta Ellis	1.8	1.10	-0.12
11.4%	Dominic McGuire	3.9	1.64	-0.18
2.0%	Klay Thompson	3.0	2.92	-0.32