Comparing coaches in the regular season (1 Viewer)

[Chase Stuart]

Let's start with a few stipulations, just to make sure the conversation doesn't go in one of the many ways I'd prefer it not to:

1. Post-season records are irrelevant. Period.

2. Making the post-season is irrelevant. Winning your division is irrelevant. The assumption is that going 16-0 one year and 0-16 the next year is equal to going 8-8 in both seasons. We're only looking at career records, which would both be 16-16 in both cases, and thus equal.

3. Assume every roster is equally talented. There is thus no better team to coach.

I'm just trying to figure out a very narrow question here, which the above stuff would just cloud. That question is how would you compare coaches with different records and different numbers of games coached?

What's better....a 75-50 career coaching record, or a 20-12 career coaching record? 200-100, or 100-50? 300-275, or 30-20? How should we weigh things to most properly rank career coaching records, exclusive of post-season berths/successes and differences in talent on rosters? Also, please eliminate the assumption that a coach that's coached 575 games must have been pretty darn good to avoid being fired. Avoid any issues of avoiding firing.

Thoughts?

 

[Chase Stuart]

Morning bump.

 

[Driver]

Let's start with a few stipulations, just to make sure the conversation doesn't go in one of the many ways I'd prefer it not to:

1. Post-season records are irrelevant. Period.

2. Making the post-season is irrelevant. Winning your division is irrelevant. The assumption is that going 16-0 one year and 0-16 the next year is equal to going 8-8 in both seasons. We're only looking at career records, which would both be 16-16 in both cases, and thus equal.

3. Assume every roster is equally talented. There is thus no better team to coach.

I'm just trying to figure out a very narrow question here, which the above stuff would just cloud. That question is how would you compare coaches with different records and different numbers of games coached?

What's better....a 75-50 career coaching record, or a 20-12 career coaching record? 200-100, or 100-50? 300-275, or 30-20? How should we weigh things to most properly rank career coaching records, exclusive of post-season berths/successes and differences in talent on rosters? Also, please eliminate the assumption that a coach that's coached 575 games must have been pretty darn good to avoid being fired. Avoid any issues of avoiding firing.

Thoughts?

I think you have to establish a minimum number of seasons coached for consideration, and then look at "adjusted winning percentage" -- like highest regular-season winning percentage, adjusted for specific factors, with a minimum of 64 games coached (or 48 or 80, etc.). Other factors that could be considered might include adjustments for (1) level of competition (strength of schedule) and (2) strength of team in year prior to coach's first season (a coach who inherits a 4-12 team and goes 24-24 in first 3 seasons is better than a coach who inherited a 12-4 team and has a 24-24 record). For example, Parcells always seemed to go to relatively poor teams that needed rebuilding. His winning percentage should be adjusted and boosted upwards compared to the guy who took over the 49ers after Bill Walsh. A simple way of doing that is to adjust the coach's number of wins/losses by the net of prior year wins versus losses, and then calculate the "adjusted winning percentage" (the 2 examples above would become 32-24 (Parcells with 24-24 record) vs. 24-32 (Walsh successor with 24-24 record)). Strength of schedule should also be easy to quantify and factor into the equation for comparison purposes.

Although it would be difficult to quantify, other factors that probably affect winning percentage between coaches are (1) draft position (most picks traded away by former regime vs. full set of picks or even extra picks waiting for new coach) and (2) salary cap situation (no cap room vs. plenty of cap room when he walks in).

 

[David Yudkin]

I think part of the issue here Chase (which serves only to muddy the waters rather than clear them) is that not all coaches inherit teams in the same situation. Would you rather get a team coming off a 1-15 season like Parcells did with the Jets or a 12-4 team like Switzer and the Cowboys?

I would think that one factor would be how many teams a guy coached. One that went 100-50 on the same team is different than one that went 100-50 coaching four teams.

As for your question, I would think the simpliest solution is bracket coaches by how many games they coached and simply go with winning percentage. Best winning % for 200 games . . . 150 games . . . 100 games, etc.

Obviously you can't go with straight winning percentage or else Chuck Drulis would be tied for the best of all-time at 2-0.

 

[Case Ace]

75 & 50 > 20 & 12 even though 20 & 12 is a slightly better %. The difference between being successful over 125 games more than makes up for being 25 % points higher over 32 games. The coach who's 20 & 12 would need either more games at his current 0.625 winning % or a better winning % over his 32 games to be as impressive as the coach who goes 75 & 50.

For the same reason, 200 & 100 is better than 100 & 50. Both are winning 67% of their games, but only one is doing it over 300 games. The guy who's only coaching 150 games would have to go something like 110 & 40 (73%) to be as good as the coach who wins two out every three for 300 games.

 

[Chase Stuart]

I think part of the issue here Chase (which serves only to muddy the waters rather than clear them) is that not all coaches inherit teams in the same situation. Would you rather get a team coming off a 1-15 season like Parcells did with the Jets or a 12-4 team like Switzer and the Cowboys?

Yes but I expressly decided to ignore that.

I would think that one factor would be how many teams a guy coached. One that went 100-50 on the same team is different than one that went 100-50 coaching four teams.

Yes, but I'd like to ignore that as well.

As for your question, I would think the simpliest solution is bracket coaches by how many games they coached and simply go with winning percentage. Best winning % for 200 games . . . 150 games . . . 100 games, etc.Obviously you can't go with straight winning percentage or else Chuck Drulis would be tied for the best of all-time at 2-0.

But how do I compare Coach A (100-80) with Coach B (20-12)? Presumably, one record is more impressive than another.

 

[Chase Stuart]

I think you have to establish a minimum number of seasons coached for consideration, and then look at "adjusted winning percentage" -- like highest regular-season winning percentage, adjusted for specific factors, with a minimum of 64 games coached (or 48 or 80, etc.).

Minimums are nice, but they don't solve all the problems. If a coach goes 16-0 in his first season, I think we can safely assume that he is a better coach than someone that goes 50-40. (Ignoring team talent, and any extraneous factors).

Other factors that could be considered might include adjustments for (1) level of competition (strength of schedule) and (2) strength of team in year prior to coach's first season (a coach who inherits a 4-12 team and goes 24-24 in first 3 seasons is better than a coach who inherited a 12-4 team and has a 24-24 record). For example, Parcells always seemed to go to relatively poor teams that needed rebuilding. His winning percentage should be adjusted and boosted upwards compared to the guy who took over the 49ers after Bill Walsh. A simple way of doing that is to adjust the coach's number of wins/losses by the net of prior year wins versus losses, and then calculate the "adjusted winning percentage" (the 2 examples above would become 32-24 (Parcells with 24-24 record) vs. 24-32 (Walsh successor with 24-24 record)). Strength of schedule should also be easy to quantify and factor into the equation for comparison purposes.Although it would be difficult to quantify, other factors that probably affect winning percentage between coaches are (1) draft position (most picks traded away by former regime vs. full set of picks or even extra picks waiting for new coach) and (2) salary cap situation (no cap room vs. plenty of cap room when he walks in).

Those are potentially useful ideas, although you'd be weighing the previous records way too heavily. But for now, I'd like to stick to just the problem at hand.

 

[Chase Stuart]

75 & 50 > 20 & 12 even though 20 & 12 is a slightly better %. The difference between being successful over 125 games more than makes up for being 25 % points higher over 32 games. The coach who's 20 & 12 would need either more games at his current 0.625 winning % or a better winning % over his 32 games to be as impressive as the coach who goes 75 & 50.For the same reason, 200 & 100 is better than 100 & 50. Both are winning 67% of their games, but only one is doing it over 300 games. The guy who's only coaching 150 games would have to go something like 110 & 40 (73%) to be as good as the coach who wins two out every three for 300 games.

Ok, this gets at what I want.You might be right that going 75-50 > 20-12, and I'd probably agree with you if I gave it much thought. The problem is how do I quantify it? How would I convince someone who disagree that we were right? And what record over 125 games would make you indifferent to a record of 20-12?Did you pull the numbers 110 and 40 out of thin air? That record seems plausible.

 

[Case Ace]

75 & 50 > 20 & 12 even though 20 & 12 is a slightly better %. The difference between being successful over 125 games more than makes up for being 25 % points higher over 32 games. The coach who's 20 & 12 would need either more games at his current 0.625 winning % or a better winning % over his 32 games to be as impressive as the coach who goes 75 & 50.For the same reason, 200 & 100 is better than 100 & 50. Both are winning 67% of their games, but only one is doing it over 300 games. The guy who's only coaching 150 games would have to go something like 110 & 40 (73%) to be as good as the coach who wins two out every three for 300 games.

Ok, this gets at what I want.You might be right that going 75-50 > 20-12, and I'd probably agree with you if I gave it much thought. The problem is how do I quantify it? How would I convince someone who disagree that we were right? And what record over 125 games would make you indifferent to a record of 20-12?Did you pull the numbers 110 and 40 out of thin air? That record seems plausible.

I did pull 110 & 40 out of thin air. I guess I'd need a formula that properly weights winning % vs. total number of games regardless of wins & losses. The variance in number of total games coached is likely to be greater that the variance in win % in most examples. What if the formula compared Win % (W%) squared to difference in games coached (G%), & then try to get that difference as close to zero as possible.110 & 40 = 150 total games @ 73.3%200 & 100 = 300 total games @ 66.7%Difference in G% = [300 - 150]/225 (average of the two totals) = 67%Difference in W% = [[73.3% - 66.7%]^2]*100 = 44%So, in this case, the difference in games (the coach who went 200 & 100) is still better. The coach who had fewer games would have to go 112 & 38 to be equal the the guy going 200 & 100 in this model.In the other example, the coach would have had to go 23 & 9 (instead of 20 & 12) to beat the other coach using this formula.