[NBA Head Coaching] By the numbers

I posted something along the lines of "if an NBA head coach witnessed and was a part of crap teams as an assistant then he's going to probably be a bad head coach, leading is own crap teams" in a discussion regarding Stephen Silas and his 20 years of assistant coaching service. @SamFisher called BS on that assumption and I truthfully never had time to compile data on coaches. Also compiling data on head coaches isn't as easy as compiling data on NBA players because it's not readily available, specifically their W-L record on teams where they were an assistant. Since I had time this week with the office being pretty slow, here is the data on the current NBA coaches (including the newly appointed Alvin Gentry):

Coach	From	To	Yrs	G	W	L	W/L%	W > .500	Asst. Coach W	Asst Coach L	Asst Coac W/L%	W/L% differential
Alvin Gentry	1995	2020	17	1105	510	595	0.462	-42.5	765	619	55%	-9.1%
Billy Donovan	2016	2022	7	489	286	203	0.585	41.5	0	0	0.0%	58.5%
Chauncey Billups	2022	2022	1	17	9	8	0.529	0.5	47	25	65.3%	-12.4%
Chris Finch	2021	2022	2	57	23	34	0.404	-5.5	378	334	53.1%	-12.7%
Doc Rivers	2000	2022	23	1713	1001	712	0.584	144.5	0	0	0.0%	58.4%
Dwane Casey	2006	2022	13	916	458	458	0.5	0	689	427	61.7%	-11.7%
Erik Spoelstra	2009	2022	14	1048	618	430	0.59	94	463	407	53.2%	5.8%
Frank Vogel	2011	2022	11	756	407	349	0.538	29	313	343	47.7%	6.1%
Gregg Popovich	1997	2022	26	1978	1314	664	0.664	325	263	229	53.5%	12.9%
Ime Udoka	2022	2022	1	17	9	8	0.529	0.5	489	230	68.0%	-15.1%
J.B. Bickerstaff	2016	2022	6	316	121	195	0.383	-37	385	583	39.8%	-1.5%
Jamahl Mosley	2022	2022	1	17	4	13	0.235	-4.5	566	631	47.3%	-23.8%
James Borrego	2015	2022	5	267	115	152	0.431	-18.5	285	273	51.1%	-8.0%
Jason Kidd	2014	2022	6	389	192	197	0.494	-2.5	94	49	65.7%	-16.3%
Luke Walton	2017	2022	6	407	166	241	0.408	-37.5	67	15	81.7%	-40.9%
Mark Daigneault	2021	2022	2	88	28	60	0.318	-16	44	28	61.1%	-29.3%
Michael Malone	2014	2022	9	596	314	282	0.527	16	460	344	57.2%	-4.5%
Mike Budenholzer	2014	2022	9	654	384	270	0.587	57	908	438	67.5%	-8.8%
Monty Williams	2011	2022	8	555	271	284	0.488	-6.5	249	243	50.6%	-1.8%
Nate McMillan	2001	2022	18	1304	696	608	0.534	44	130	106	55.1%	-1.7%
Nick Nurse	2019	2022	4	244	146	98	0.598	24	263	147	64.1%	-4.3%
Quin Snyder	2015	2022	8	570	334	236	0.586	49	161	151	51.6%	7.0%
Rick Carlisle	2002	2022	20	1543	843	700	0.546	71.5	406	382	51.5%	3.1%
Stephen Silas	2021	2022	2	88	18	70	0.205	-26	729	888	45.1%	-24.6%
Steve Kerr	2015	2022	8	564	391	173	0.693	109	0	0	0.0%	69.3%
Steve Nash	2021	2022	2	89	60	29	0.674	15.5	0	0	0.0%	67.4%
Taylor Jenkins	2020	2022	3	161	80	81	0.497	-0.5	273	219	55.5%	-5.8%
Tom Thibodeau	2011	2022	10	687	402	285	0.585	58.5	849	759	52.8%	5.7%
Tyronn Lue	2016	2022	6	300	185	115	0.617	35	186	115	61.8%	-0.1%
Wes Unseld	2022	2022	1	16	11	5	0.688	3	551	726	43.1%	25.7%
Willie Green	2022	2022	1	18	3	15	0.167	-6	142	85	62.6%	-45.9%

Now does my previously statement hold true, that if you sat on that bench as an assistant with crappy teams you'd eventually lead your own crappy team as a head coach? Well let's take a look.

1. The first thing I want to call out are the ZACs, Zero Assistant Coaching. This group is comprised of Steve Kerr, Steve Nash, Doc Rivers and Billy Donovan. If you were wondering about Billups he had one year with the Clippers. Anyhow, these four guys have a cumulative win total of 1738 wins against 1117 losses, good for a 60.9% winning percentage. This is easily better than the remaining 27 head coaches' cumulative percentage of 48.6%, or almost a +13% difference. I am calling the ZACs out not because they are winning at a pretty high margin (60.9% is about a 50-win season) but because the statement I previously made was exclusive to head coaches with assistant coaching experience. So for this exercise I'm removing the ZACs.

2. Once I remove the ZACs, we are left with 27 head coaches (due to the recent firing and appointment of Walton and Gentry, respectively). In terms of wins and losses, there's a wide range in terms of head coaching experience as you can see above: some have only had a year or two as an assistant like Billups or Daigneault whereas there have been "seasoned" HCs with years of assistant experiencing like Thibs, Budenholzer and our very own Steve. This next call out might ruffle some feathers...some might even think I'm cherry picking data to fit my own narrative. Well that may be but it's not an egregious removal. What I want to remove are the head coaches who only have 1 or 2 years of experience at the lead gig. The Silas's, the Bilups's, the Udoka's, etc of the coaching world can be excluded because their data has simply too much noise (e.g. W/L% differential between their HC record vs their assistant coaching record). So 8 coaches are removed: Billups, Green, Udoka, Unseld Jr, Mosley, Daigneault, Finch and Silas.

 

3. So now we're left with 19 head coaches (including poor ole Luke). Across these 19 coaches there are 13796 victories, 7556 losses, and 199 years of cumulative coaching experience.

Coach	From	To	Yrs	G	W	L	W/L%	W > .500	Asst. Coach W	Asst Coach L	Asst Coac W/L%	W/L% differential
Taylor Jenkins	2020	2022	3	161	80	81	0.497	-0.5	273	219	55.5%	-5.8%
Nick Nurse	2019	2022	4	244	146	98	0.598	24	263	147	64.1%	-4.3%
James Borrego	2015	2022	5	267	115	152	0.431	-18.5	285	273	51.1%	-8.0%
Luke Walton	2017	2022	6	407	166	241	0.408	-37.5	67	15	81.7%	-40.9%
Jason Kidd	2014	2022	6	389	192	197	0.494	-2.5	94	49	65.7%	-16.3%
Tyronn Lue	2016	2022	6	300	185	115	0.617	35	186	115	61.8%	-0.1%
J.B. Bickerstaff	2016	2022	6	316	121	195	0.383	-37	385	583	39.8%	-1.5%
Quin Snyder	2015	2022	8	570	334	236	0.586	49	161	151	51.6%	7.0%
Monty Williams	2011	2022	8	555	271	284	0.488	-6.5	249	243	50.6%	-1.8%
Michael Malone	2014	2022	9	596	314	282	0.527	16	460	344	57.2%	-4.5%
Mike Budenholzer	2014	2022	9	654	384	270	0.587	57	908	438	67.5%	-8.8%
Tom Thibodeau	2011	2022	10	687	402	285	0.585	58.5	849	759	52.8%	5.7%
Frank Vogel	2011	2022	11	756	407	349	0.538	29	313	343	47.7%	6.1%
Dwane Casey	2006	2022	13	916	458	458	0.5	0	689	427	61.7%	-11.7%
Erik Spoelstra	2009	2022	14	1048	618	430	0.59	94	463	407	53.2%	5.8%
Alvin Gentry	1995	2020	17	1105	510	595	0.462	-42.5	765	619	55%	-9.1%
Nate McMillan	2001	2022	18	1304	696	608	0.534	44	130	106	55.1%	-1.7%
Rick Carlisle	2002	2022	20	1543	843	700	0.546	71.5	406	382	51.5%	3.1%
Gregg Popovich	1997	2022	26	1978	1314	664	0.664	325	263	229	53.5%	12.9%

Looking specifically at the win differential (head coaching vs. assistant coaching) if we take the absolute value of each coach's % and generate an average across 19 coaches, we have a +/- 8.2%. In terms of NBA regular season games, 8.2% is about 7 games. That's a pretty good margin: in general terms, an assistant coach taking over a team that went 41-41 the season prior could, theoretically, have that team anywhere between 34 and 48 wins if every other variable remains constant from the previous season.

Now, do I go ahead and try to "cherry pick" my data even more? Sure, let's do that and the first thing I'm looking at is Walton. Luke's numbers are heavily influenced by two extremes: his two years as an assistant on historic Warriors teams and his actual mediocre head coaching gig. "But Sted, didn't you want to prove that a guy having a great W/L record as an assistant would tend to yield a similar W/L record as a head coach? By you eliminating Walton you're not only being hypocritical but will also unnecessarily feed the trolls!". Well, for starters, based on the W/L % differential alone, Walton can/should be considered a statistical anomaly. Second, we all know the Warriors, by now, can function without a head coach. With both of those reasons, I'm going to wipe Luke off the list. By doing that the +/- shrinks from 8.2% down to 6.3%, or +/- 5 games. There are still extreme cases, where the % differential > 10% (like Pop with a +12.9% or Kidd with a -16.3%) but paring down the data even more wouldn't move the needle (6.3% down to 4.9% when removing Pop, Kidd, and Casey) as it did by just removing Walton from the equation.

So, what's the conclusion? I guess what the data says, at the surface, is that a head coach's Win-Loss record wouldn't deviate too much from what he saw while riding pine as an assistant. JBB has a lifetime 38.3% winning percentage has a head coach and a 39.8% winning percentage as an assistant over almost 1000 games. Mike Malone's head coaching record is only about 4% lower (57.2% vs 52.7%) than what he saw as an assistant. So where does that leave us for the new/recently new head coaches?

 

Coach	From	To	Yrs	G	W	L	W/L%	W > .500	Asst. Coach W	Asst Coach L	Asst Coac W/L%	W/L% differential
Chauncey Billups	2022	2022	1	17	9	8	0.529	0.5	47	25	65.3%	-12.4%
Willie Green	2022	2022	1	18	3	15	0.167	-6	142	85	62.6%	-45.9%
Ime Udoka	2022	2022	1	17	9	8	0.529	0.5	489	230	68.0%	-15.1%
Wes Unseld	2022	2022	1	16	11	5	0.688	3	551	726	43.1%	25.7%
Jamahl Mosley	2022	2022	1	17	4	13	0.235	-4.5	566	631	47.3%	-23.8%
Mark Daigneault	2021	2022	2	88	28	60	0.318	-16	44	28	61.1%	-29.3%
Chris Finch	2021	2022	2	57	23	34	0.404	-5.5	378	334	53.1%	-12.7%
Stephen Silas	2021	2022	2	88	18	70	0.205	-26	729	888	45.1%	-24.6%

Some of these guys (Billups, Green, Udoka) came into their respective HC gigs with very high success from their time as an assistant. So far, out of the three, two are barely over .500 and Willie Green has a mess in New Orleans. On the other end of the spectrum, Wes Unseld Jr. has, so far, performed better (+25.7% differential) than his assistant record states.

"A-ha! So you're saying that assistant coaching record has ZERO correlation to how a guy would do as a head coach! Which means you're admitting that Silas has been wrongfully blamed for the Rockets' lack of success and should be given more time!!!"

Mmmmm no, not saying that at all. What I am saying is that while Silas might not be as bad as his current head coaching record appears (looking at his small sample size of 88 games coached) , the trend among assistant coach to head coach tell us that he doesn't have a lot of run way to work with given his winning % (45.1%) as an assistant over 1600 games. If we apply the +/- % (6.3%) from the 18 coaches you're looking at a head coach with a potential min of 38.8% (or 32-win season) or 51.4% (42 wins). Of the 18 coaches that are left (once we remove the ZACs, the newbies, and Walton), 75% of the list has performed worse as a head coach than they did as an assistant based on W/L%. The 6 that DID perform better has a HC vs their time as an assistant are coaching or have coached some of the best teams (e.g. championship) in the NBA right now: Carlisle, Thibs, Spoelstra, Vogel, Snyder, Pops. Of course most of these coaches also benefitted from having transcendent players on their roster (Carlisle had Dirk, Spoels had the Big 3, Pops had his own Big 3, Vogel had LeBron) so they could have very well been closer to a +/- of 0% than what their + % differential says.

TL;DR: Picking an assistant who's entire career has seen success doesn't automatically translate into his own success as a head coach (+/-6.3% between his assistant W/L record vs his HC record) but starting out with a higher winning percentage as an assistant definitely doesn't hurt.

Pick through, analyze, criticize, etc. whatever you want. Just thought I pass along some data.

 

Fun facts:
Of the 31 (including Walton) head coaches, 12 were former NBA players. Of those 12, 3 were forwards (Walton, Williams, Udoka), 2 were shooting guards (Green, Carlisle) and 7 were point guards. Their cumulative winning % as head coaches is 55.3%

Stephen Silas has the most assistant coaching experience among the 31 coaches, with 1617 games. Second on that list is Thibs.

With Walton canned, the current head coach with the highest winning % as an assistant is Ime Udoka with 68% and second is Budenholzer with a 67.5% winning percentage. Not surprisingly, both came from the Spurs. FWIW, Walton's % as an assistant is an absurd 81.7%