There are exceptions, like that guy on the left, but my hunch is that NFL coaches are mostly interchangeable.
I think at the NFL level, all coaches employ the same best practices. There is no secret sauce that one coach has over another in terms of instruction, motivation, strategy, etc. This is because of the highly mobile, fluid market for coaches and the large size of their staffs. There are very strong constraints on deviation from league norms in any dimension.
Also, from statistical analysis, we can measure the variance in team performance attributable to randomness (sample error due to a short 16-game season) and player impacts (the addition or subtraction of a player's impact on team production, player interaction effects). There is very little variance left that can be attributed to other causes, including coaching. In other words NFL outcomes are overwhelmingly driven by player talent and luck, and there's not much room left for coaching to make a big impact.
We can observe this intuitively, as the very same coaches can have wildly different records from year to year. How much effect can they really have?
That said, if you made a clown like me head coach of the Ravens last year, I guarantee they would not have won the SB. So coaching obviously matters. It's just that NFL coaches are more or less equal (equally excellent), for the reasons stated above. In a competition system where participants are roughly equivalent, other factors will dominate the outcome. You can still really screw things up if you don't know what you're doing, but it's exceptionally difficult to stand out as superior.
Consider a simplified analogy. Rock Paper Scissors is a game where there is skill involved. The skill is: don't be predictable. Once everyone has that skill, outcomes are determined by other factors--in the case of RPS, it's randomness. It would be very easy to lose on purpose by acting unskilled but very hard to win on purpose, which suggests skill is involved. But if everyone is on the same skill level, skill becomes irrelevant.
I'm not saying that NFL coaches with poor records are losing on purpose. To be clear,I'm suggesting is that it's much easier for coaches to be on the downside of the equation than on the upside of the equation. There is an asymmetry to the top end of the "coaching function." In graphical form it might look like this:
If the x axis is coaching skill, then all NFL coaches, even the relatively "poor" ones, are going to come from the right side of the graph. Consider the y axis to be "effect of coaching on team outcomes." In the top echelons of coaching, there are decreasing returns to better and better coaching. As you move right along the curve, there is less and and less room for positive effect. But if you move leftward, the slope is steeper, which means it's easy to have a negative effect, but hard to have a positive effect as a coach.
(As an aside, this curve may explain why there is so little risk taking among NFL coaches. The perceived upside is smaller than the perceived downside. In most cases, this may be true.)
I realize this is not a widely accepted view. People attribute group success and failure to leaders for many reasons. First, it's in our nature. The fundamental attribution error causes people to put blame for failures on others but claim to be the cause of success. So the coaches who happen to lead winning teams are the first to perpetuate the myth of great coaching. The second group to perpetuate the myth are GMs and owners who fire coaches who happen to have losing records. Everyone is fooled that changing coaches has a larger effect on team outcomes because of regression to the mean. A team that wins 4 games one season is bound to improve next season for all kinds of reasons not related to the coach. Firing and replacing a losing coach doesn't appear to have much effect beyond what regression tells us would happen anyway.
The second factor is the illusion of control, in which people overestimate their own ability to control outcomes in life, good and bad. We naturally project this overconfidence onto others, believing they have similar levels of control over events. But how much control did John Harbaugh have over Billy Cundiff's missed 32-yd FG in the 2011 season's AFC Championship Game? Or the now-legendary misplay by Denver safety Rahim Moore in the game-tying bomb in the playoffs last season?
Third, there is a philosophical and even ideological bias toward attributing outcomes to great leaders. The prevailing ideology in academia and the media sees history as a long line of great leaders whose personal vision and actions drive world events. Who is credited with freeing the slaves? Is it the millions of soldiers who fought for the Union in the Civil War? Is it the millions of people who participated in the emancipation movement? No, it's one guy--Abraham Lincoln. The other side of the philosophical debate sees events as caused more from the bottom up--societal trends drive history. And the other side of the debate sees events as more caused by the top down. And it's the top-down worldview that currently reigns.
This narrative is personified in Tony Dungy. Dungy is a beloved former coach and champion. I consider myself a fan of his. But remember the story when he took over the Colts and finally won a Super Bowl for Peyton Manning? The narrative was that the Colts had great players but couldn't get it done until they had a great leader like Dungy to put them over the top. But there was also the matter of Dungy's previous job in Tampa Bay. The Buccaneers won the Super Bowl the year after replacing Dungy. The narrative then was, 'Well, they did it with Dungy's players.' In one case, it's the players that makes the difference. In the other case, it's the coach that makes the difference. Both narratives are diametrically at odds, yet live on in the very same person.
Another factor is an illusion generated by network effects at the college level. Great high school players and junior college transfers understandably congregate toward winning college teams, perpetuating and enhancing the superiority of those programs. The coaches at the top schools naturally gain stature, which in turn further perpetuates their ability to recruit top talent. It's a self-reinforcing, rich-get-richer system. Once a coach has enough recognition, he becomes a franchise unto himself.
Consider Nick Saban, who is widely regarded as an unrivaled great football coach. His pro team was decidedly mediocre in Miami, and it was worse his second year than his first. One explanation for the difference is that he happens to have some sort of mysterious knack for college coaching but not for pro coaching. But no one can put a finger on what it is that is so qualitatively different about the function of a coach at either level. The alternative theory is that coaching doesn't have a large effect in either case, and that it's the players that matter. And Saban simply has a huge advantage in player talent at Alabama that he could never have in Miami.
Some claim there are large differences among coaches in the skill of player evaluation. How often are we reminded of how Bill Belichick drafted Tom Brady in the 6th round of the draft? But we rarely hear about how he's not able to draft defensive backs. We hear the name Jeff Marriott even less often. Who's Jeff Marriott you ask? He's the defensive tackle New England chose with its 5th round pick the same year they took Brady. Marriott never played a snap in the NFL. If Belichick really had any inkling at all that Brady would be a solid QB, much less a Hall of Fame bound superstar, don't you think he might have used a measly 5th round pick to grab him, just to be safe? The truth is he thought Jeff Marriott wasn't worth risking leaving on the board until the 6th round, but Tom Brady was.
Don't get me wrong. I'm not saying Belichick is a poor talent evaluator. I'm claiming he's indistinguishable from most other coaches and GMs in that department. He has hits and misses just like everyone else. But because of his team's success, he has been bestowed an aura of greatness.
The importance of the coach is overrated in the NFL for a number of reasons--mathematical, statistical, psychological, and philosophical.
Subscribe to:
Post Comments (Atom)
I believe Leo Tolstoy write 'War and Peace' in an effort to disprove the 'Leadership Principal' that was perpetuated during and after the reign of Napoleon. I always imagined that Napoleon would coach like Belichick or Sean Payton. Very interesting article.
Cheers,
J
A lot of words, yet zero evidence supplied that coaching doesn't matter in the NFL. Great conclusion as well: Belichick didn't draft Brady early; therefore, he is average evaluator of player talent/value.
You should try following some of the links in the article.
I don't know of any reason to disagree with this. But as a Saints fan (a lot happier this century than last), may I just say stress the fact there have been and are some truly dumb coaches in the League who definitely impact, if not the Ws and Ls, at least the nature of the performance and quality of their teams' game. Some instances of salient impact: 1. insane draft picks (Saints wasting a #1 pick on mediocre kicker Russell Erxleben; Jets picking whoever it was instead of Warren Sapp--both these were driven mainly or largely by the head coach, I believe); 2. bone-headed, to all appearances inexplicable strategic decisions, e.g. team running rampant over the opposition in 1st half suddenly going all-pass in 2nd half (Ditka; also, a few games every season), or repeated loss of either poise, or mind, in the 4th quarter. Can't help but suspect a good number of the coaches, this year and each past season, are somewhere along the steeper part of that slope.
"Great conclusion as well: Belichick didn't draft Brady early; therefore, he is average evaluator of player talent/value."
I believe the idea was to refute a specific piece of evidence bandied about by the mainstream media folks to support Belichik's bona fides as THE supreme talent evaluator in the NFL. The Patriots passed on Brady five times that draft, same as most teams. And then, after drafting him, didn't even start him until Drew Bledsoe was hospitalized. The very fact that Mo Lewis changed the course of the NFL with a single hit means that the Patriots had little idea they were sitting on a freaking gold mine.
The article's point is that Tom Brady is a terrible example to support the assumption that the Patriots are excellent evaluators of talent. A better example would be a player who was "reached for" that turned out to be just as fantastic a player (or close to it).
> A lot of words, yet zero evidence supplied that coaching doesn't matter
> in the NFL.
Brian clearly believes that coaching matters:
"...you made a clown like me head coach of the Ravens last year, I guarantee they would not have won the SB. So coaching obviously matters."
Rather, the claim is that the great NFL coaches like Belichick and Dungy probably aren't that much better at winning than less celebrated ones like the Harbaugh brothers. (People do suggest that there's about an expected game of difference in the better and worse 4th down decisions that coaches make.)
As for talent evaluation - it's not just Brady or Belichick. Cade Massey published a paper suggesting very strongly that all NFL teams are about equally good at talent evaluation.
I do agree coaches tend to gather around a mean and you could probably do a musical chairs of 20-25 coaches and see only a variation of +/- 2 games but the difference between a very bad and very good coach is broad, as noted with the Chiefs. The only real change was Alex Smith and Andy Reid. Smith hasn't really done a whole lot but the team has already tripled it's win total from last year, and it's because they went from a very bad coach to a very good one. So while on average coaches are interchangeable, there is a difference in the extremes, which we see in Kansas City.
Maybe you could compare coaches actual decisions to the WPA model for fourth down decisions to see how much effect an optimal coach could impact the team over the course of a season.
Topher Doll,
That's simplifying the matter quite a bit. While the consensus is that Romeo Crennel is a pretty awful coach (and thus, one would expect that an improvement in coach would have a noticeable impact), attributing all of KC's improvement to the coaching change ignores everything else. While the Chiefs were awful in terms of win-loss record last year, they already had a number of good players whose impacts were counteracted by some awful players, and it's because there were some awful players that improving on them would improve the team significantly. Alex Smith hasn't been great, but he's still been significantly better than what they had last year, if only because his turnovers are drastically lower (and while one should expect Alex Smith to be a low turnover QB, Cassel and Quinn also performed even worse than what might be expected from their talent, which means the Chiefs would likely have a better record if they simply replayed last season with the same players). There have been significant improvements from the players on defense already there, most noticeably Dontari Poe, and they added Sean Smith as well. If anything, the reason the Chiefs are significantly better this year is their number 1 ranked defense while their offense is still in the 20s, and Andy Reid is, if anything, known as an offensive guy. I don't doubt Reid might have some leadership qualities that would impact the team as a whole (even the units he's not in charge of) and Crennel may have prevented the players from reaching their full potential last year, but it's much more likely that the majority of the Chiefs' improvement has been from other factors.
Great post Brian. I have a few thoughts and questions:
1. On risk-taking, I agree that coaches tend to be risk-adverse, but this is perhaps mitigated (exacerbated) by risky behavior that they think is non-risky. The fourth-down issue is the most obvious example, but there are others. These raise the question of how to evaluate someone who claims to be risk-adverse (or otherwise acts in a risk-averse way) but then (presumably unknowingly) engages is risky behavior.
2. Two points on the psychology/great leader analysis. First, your point about top-down versus bottom-up is, I think, a little off-point. Even if you believe change comes from the bottom up, the timing of change is often (arguably, by definition, always) driven by individual actors and therefor the personalities of those actors is important in when the change occurs. Football games are discrete events so that the timing of change is particularly important for these events. Second, and maybe this is just an element of your analysis that there are some lousy coaches, there does appear to be an identifiable skill set for head coaches as opposed to position coaches or coordinators (consider, for example, Norv Turner).
3. I think I agree that personnel decisions are more important than the head coach. That makes we wonder whether we can identify particular teams or people who are better at this and construct a sort of "futures" rating of teams.
4. Finally, slightly off pint, but stemming from the personnel point above and an issue I heard Barnwell discuss on his podcast, I wonder whether GMs or coaches or schemes have injury impact. Are some GMs better at evaluating injury risk in players? Are some coaches better at managing practice and conditioning to limit injuries? Is there any correlation between types of plays and injury occurrence?
Keep up the good work.
Re: Tony Dungy - I have a hard time thinking Dungy is a great coach. Some of the things he says on Sunday night football seem totally backwards and/or arbitrary. Even on replays where he supposedly put a lot of thought into something. Can't think of examples right now, but Chris Collinsworth seems much more in tune with the flow of a particular game, and Colinsworth is just talking off the cuff. Maybe Dungy is a good coach because he is a straight forward and honest guy, and people like that.
"NFL outcomes are overwhelmingly driven by player talent and luck, and there's not much room left for coaching to make a big impact"
Except for the intangible ability for a coach to inspire his players. Witness teams that have given up on their coaches, like last year's Cardinals, and as a counterpoint, the Colts rallying around Chuck Pagano. I think those coaches who don't have the magic of inspiration are the ones who don't make it. A toxic relationship between the boss and the workers results in players that are just putting in their hours and collecting a paycheck.
This post reminds me of Daniel Kahneman's quote that "Finance is an entire industry built on the illusion of skill."
Brian, you've pointed out elsewhere that mean regression is the name of the game in the NFL. (Notable exceptions being the Patriots and Lions over the last 10 years or so.) That fact and this post seem to cast a pall over the other features of your blog--isn't it all randomness? Why bother measuring efficiency and such? If you put a bunch of evenly-matched teams into a monte carlo simulation you're going to get normally distributed, random outcomes. It would LOOK like there were differences in talent and coaching as a result. That's increasingly how I see the NFL. We don't want to see that it's random, we want to believe that our favorite players and coaches are so awesome that they turned around a 3-13 team the previous year to an 11-5 powerhouse the next. But it sure is hard to prove that it wasn't due to randomness.
I'm not sure if anyone has brought this up, but the Tom Brady example is good in spirit but ultimately a bit flawed.
Who's to say that Bill didn't think he was going to be a 7th round pick and took him in the 6th just to be sure?
And if he had taken him in the 5th, why not the 4th? It's very clear now that no one should have been picked before Tom Brady, and this was a draft with 14 first round picks who would go on to be Pro Bowlers (granted, some are of the Bubba Franks and Sebastian Jankikowski variety).
Anyway, very good and interesting article. Well worth the read.
Same poster as above, just wanted to add one more thing - the Patriots themselves even passed on Tom Brady in the 6th round, choosing Antwan Harris, a DB out of Virginia, 187th overall.
I know you might say it's splitting hairs here, but I think Mr. Brady has had a slightly better career.
I very much agree with you, in general.
I'm old enough to remember when NFL coaches were hired because they were drinking buddies of the owner or "names" who would sell a few more tickets (Sammy Baugh, Otto Graham) when that really mattered. In those days the difference between the top and bottom levels of the coaching ranks was big and very real. But when the media money started pouring in the owners realized they were losing a lot of potential revenue with bad coaching and started moving the bottom level up, with the less capable candidates purged on the way to the top job. Now the quality differences among coaches is much much smaller.
But "coaching doesn't matter" is a very false conclusion to derive from a belief that all NFL coaches produce pretty much the same results (even if true). Dave Berri has said that about NBA coaching, I don't know if it is just sloppiness on his part or if he really believes it, and a lot of people repeat that in Berri-land. But the fact that most all coaches are about equally excellent doesn't mean getting a less good coach won't kill you.
The same naive argument is often made about corporate CEOs: since none have a clear ability to beat the market they are all grossly over-paid and anyone could do the job. This misses the point that CEOS are paid largely to not lose huge amounts of money -- firms pay to get a guy who is good enough to not get them wiped out in an extremely competitive environment. A bad CEO can destroy billions of dollars very quickly (Enron, Worldcom, what the new guy just did to JC Penney in mere months).
In econ and evolutionary science there is a concept called Red Queen Competition that applies when competitors all have access to the same resources and information and can copy each other. Then the best of the best can only stay together in a pack (one may get a temporary advantage over the rest but there is no way to make it permanent) while the less than the best drop back to the point of being exterminated.
With revenue sharing, salary cap, free agency, I imagine I see signs of RQC growing throughout the NFL. E.g. as over a decade the pack of top teams divvy up playoff wins and rings, the Millen Lions flatline dead on the bottom (though league rules prevented them from actually being exterminated and now being extinct). I think one can see this in position play and on other smaller scales as well. (If real life ever gives me a break, I've got some ideas for a couple Community posts on this.)
Indeed, leadership is interesting. Gneisenau, Scharnhorst, and Clausewitz created the Prussian General Staff so that the genius of Napoleon could be re-created by a committee. In the four major wars fought by that institution they are 2-2 but the Prussian/German army has cleary been far more efficient than its opponents; they have consistently inflicted higher casualties upon their opponents than they themselves have suffered. In football, finance, and war victory usually goes to the least incompetent.
Cheers,
J
Of course! I should have linked to Cade's research on draft and luck.
"NFL outcomes are overwhelmingly driven by player talent and luck, and there's not much room left for coaching to make a big impact"
While I generally agree with you, as per my prior comment, ISTM this overstates things and overlooks ways in which coaches do still significantly make a difference.
[] First, this is dangerously close to saying "coaching doesn't matter", when as per the RQC model coaching matters *a lot*, while the similar results obtained by coaches come from their similar levels of very high skill, with the less good coaches culled out on the way to the top job or if they get there very quickly then. (E.g., I personally think Singletary was clearly a less good coach, not just an unlucky one, and was culled quickly for being so.) So the culled coaches don't have much effect on the data. A sample selection issue.
Hmmm ... it's kinda like the argument I've seen that SAT scores "don't matter" because they don't predict student academic achievement, based on the scores of students in Ivy League Schools who are all in the top 2%. Compare the scores of students in Ivy League students to those in community colleges and try to say that.
[] I certainly agree that the knowledge and "technical" skills of HCs today are far more balanced at a much higher level than in past decades. Not much difference there. But there's much more to a leadership position than that.
In my experience in life and in watching pro coaches, *dealing with other people* is often very much the toughest part of the job, often the most important, no matter the technical skills required. I've seen coaches who were genuinely good elsewhere come to NYC and implode dealing with the press and fans (Coslet) and clearly coaches can bungle dealing with their own players in ways that hurt the team (Singletary). That mattered.
Being the right person for the job -- fitting well and complimenting all the other people and circumstances, instead of causing wheels to grind and being hated -- matters a lot. The shape of your peg had better fit that of the hole you are trying to fit into, so to speak. GM's aren't necessarily stupid when the replace a disciplinarian coach with a players coach, and vice versa. Now, one can have a philosophical argument about whether landing in a job you are right for instead of one you are wrong for is "skill" or "chance", but it matters.
[] Plainly, the organizational quality of a franchise matters a lot -- the Pats and Steelers being at the top and Lions at the bottom for a full decade was not mere luck.
Primarl responsibility for this has to drop on the owner, but as to actual operations and managing their quality the HC has a heck of a lot of input. Maybe the GM has more, or maybe not. But who else has as much?
E.g., "outcomes are overwhelmingly driven by player talent", sure, but the HC has a significant say in selecting that talent and 100% responsibility for developing it. If he does a *great* job at building up the talent, by the logic here he gets no credit for the resulting Ws because the talent won the games, not the coaching.
So ISTM that you are really saying "game day decisions by the HC don't matter near as much as fans believe", with which I'd agree 100%. (Though there a whole lot of posts on this blog about their importance.) But there's a lot more to coaching than that.
Heck, Lombardi said the real work of a coach building a team was done in the offseason, then in preseason, then in planning before a game -- so by Friday it was over and he was relaxing, and during the game he felt like a cheer-leading fan with little else to do. So by that game-day measure, he'd have agreed that "coaching doesn't matter" too!
You wrote " But no one can put a finger on what it is that is so qualitatively different about the function of a coach at either level." [college or pro]."
Not so sure. Mack Brown left North Carolina for Texas. In his first year in Texas, before the first game, he met every high school coach in the state of Texas. As you point out, recruiting is a complex subject, and there are other factors besides the coach. But I think recruiting is a major factor in major college football coaching, and not much of a factor in pro football coaching. That is a difference in the nature of the jobs.
Couldn't all that's being said about coach's, also be said about quarterbacks? They are all 99% the same. They are experts at what they do, they have have gone through an exhaustive vetting process to get to their position, and they are all running more or less the same offenses. Further, being an elite quarterback is neither necessary nor sufficient for team success (Trent Dilfer in 2000 and Drew Brees last year).
The difference is, we have reliable ways to quantify a QB's influence on a game, but not a coach's.
I guess I take issue with the assertion that there's some upper limit of coaching ability, beyond which all differences are negligible. Football isn't rock, paper, scissors; it's chess.
Brian,
Can you discuss (or point to) the research you've done on "player impacts" that you are saying accounts for the remaining variance in outcomes after sampling error?
Thanks!
I think the chess analogy is interesting. Opposing coaches could be considered the chess-masters playing one another. Except that the pieces fight with each other and actually determine which piece takes which square. There's only so much a chess-master can do, even a grandmaster.
Same anon as above,
I think the chess analogy is interesting. Opposing coaches could be considered the chess-masters playing one another. Except that the pieces fight with each other and actually determine which piece takes which square. There's only so much a chess-master can do, even a grandmaster.
Granted it's an imperfect analogy, but for the sake of continuing the discussion, suppose a great coach/chess-master can consistently manufacture scenarios in which his queen is fighting his opponents pawns for desired squares, while keeping his own pawns out of high-leverage fights. That is, exploiting weaknesses of opponents, while mitigating his own.
I think your scenario assumes that on any given "turn", there is some definite optimum move that any NFL coach would know to make (which square to attack), and so it's just up to the pieces to execute that move. The problem is, with few exceptions, unless you simplify a coach's options to run, pass, punt, kick, there are no definite optimal moves on any given play, since any decision is to some degree dependent on what you anticipate your opponent's counter-strategy will be.
Now, if a great coach has nothing but pawns, he'll be out-matched regardless. But in an all-else-being-equal scenario, his ability to manipulate his pieces will surely influence a match's outcome.
I couldn't disagree more with the premise of the column, and I don't find the provided support to be in any way compelling. It's not even circumstantial so much as hypothetical. To me, this seems like another misguided attempt to bring a baseball concept (the manager not being important) into football, much like some insisting that records in 7 point games being luck (inaccurately assuming they correspond to 1 run games).
I'm a huge stats guy when it comes to baseball, but the difficulty in applying similar analysis to football is that it so rarely involves discrete, independent events. Did Sean Payton's absence have a significant effect on New Orleans' season in 2012? Undoubtedly, and yet it's hard to isolate the degree of that effect from other factors. But what we can recognize when it comes to football is that head coaches have a much more direct impact on the game than manages in baseball. There is no "play-calling" in baseball, and there is far less in-season training as well due to playing nearly every day. The much smaller sample of NFL games also means that decisions in key situations have a greater impact on the season than a similar number would have amidst a 162 game schedule.
If simply having "the best players" corresponded to NFL success then the Kansas City Chiefs wouldn't have won two games last season with six Pro Bowlers (I'm aware that the Pro Bowl is a flawed measure, but everyone agrees that KC had abundant talent). Anyone who follows football closely recognizes that talent alone isn't sufficient. In fact, it's more about execution than talent, and if we recognize that then we also must recognize that while coaches have no effect on talent they would have a considerable effect on execution both in terms of what players are asked to do and how well they have been trained to do it.
And regarding the talent evaluation angle I will make two points: #1) Not identifying every player accurately (in this case Brady) and not identifying players accurately more often than peers are two wildly different things. #2) More importantly, if analysis at a certain level of indistinguishable then people have absolutely no reason to come to this site since your argument is that some other site on the internet with a similar background would be just as good.
Has anyone bothered to run a regression that looks at, say, team record after a coaching change? What happens if you lag the "new coach" variable a couple seasons?
I think the chess analogy is interesting. Opposing coaches could be considered the chess-masters playing one another...There's only so much a chess-master can do,
Only if one considers head coaching to be equal to play calling. I'd say game day play calling is only about 10% of head coaching -- and to the extent that in today's NFL plays are called by OCs and DCs, 0% of head coaching.
I was a tournament chess player for some years. That was 100% "play calling", 0% developing the highest-quality fighting pieces to win games for me, by replacing, teaching, developing them. Pretty big difference, IMHO.
All the great coaches I've ever seen talk or write on the subject have said "coaching is teaching", I've never seen one say "coaching is play calling". But to the extent such teaching improves the players, the logic here gives the coach gets no credit for it, because the wins come from having the better players, and how the coach got better players doesn't count.
I'd think most people would agree that the job of a head coach includes talent evaluation, roster management, developing individual players in their position play, developing a team system that best uses their particular talents together, teaching that system so it is understood and drilling the players so they "don't think" when playing it, motivation, human relations, in today's NFL effectively being a CEO (or at least COO) directing many other people and developing high-quality assistants and subordinates, weekly game planning, and game-day play calling.
All those tasks but the last two feed into the "quality of the players" to a greater or lesser extent -- and with the rise of OCs and DCs, coaches now often don't even do the last two.
I completely 100% agree that coaches today don't make as much difference between teams as many people imagine or as they used to (by the RQC model, not any "coaches don't matter" notion.)
I don't agree at all with the logic that 'the best players and random chance so fully explain winning that it doesn't leave much room for any effect of coaching' -- as 'the best players wins but for luck' is pretty much a tautology.
If you applied the same statistical analysis to the teams of Lombardi in the 1960s and Paul Brown in the 1950s you'd reach the same statistical conclusion -- they won because they had the best players, coaching had very little to do with it, year after year after year ... But how did they have the best players?
My main question is, how do you explain the continued dominance of the Patriots for the entire Belichick tenure (excepting the first year), despite almost 100% turnover?
To the extent that team performance is consistent across complete or near-complete player turnover, which I'm fairly confident is true for some teams both good and bad, what do you ascribe it to? Purely general managers choosing players correctly or incorrectly? It seems, moreover, that there are many examples of successful teams (Patriots, Saints, Steelers) slurping unsuccessful players from unsuccessful teams and turning them into hugely successful and apparently very key players. Do you think this is purely random?
Tom Brady.
The only evidence you offer is that most outcomes are determined by chance and player performance. But player performance could itself be affected by coaching. In fact, I would expect this to be precisely the mechanism: a good coach makes his players play better.
A simple test would be to look at the variance in the change of record from one year to the next of two sets of teams: those with the same coach in both years, and those that had a coaching change.
I like the chess analogy in that a grand master (most NFL coaches) beats another grand master by the gradual accumulation of small advantages ( according to Alekhine ), and the elimination of mistakes.
Have any of you read Coaching Matters: Leadership and Tactics of the NFL's Ten Greatest Coaches? The best NFL coaches seem to win almost every year, no matter the talent or circumstances they inherit. I wonder why?
In regard to the "Tom Brady" response to the Patriots' question, I guess Belichick just got lucky with superstar Matt Cassel.