foosballnick wrote:nixluva wrote:foosballnick wrote:nixluva wrote:foosballnick wrote:I remain dubious about any analysis measuring the effect of Change Agents (in this case Head Coaches) without understanding how the model that was created accounts for a Constant (in this case normalizing factor for impact of Roster Changes on Head Coaches success). In other words - it is difficult to measure the impact of a coach without understanding the impact of talent & changes to the roster and competition from year to year. To use the referenced blog as a resource to statistically prove that coaches have little to no impact would make a true Statistician's head spin.
Sounds like you should come up with more data points and a set of criteria to add to what has already been done. Having read the few attempts that were made I was fairly convinced that it was close enough to make the case. Just based on what we know anecdotally Pop and Phil would be considered to have the greatest impact. Very few other examples come to mind. Thibs might be a coach who I could believe would have an impact but just how much of an impact? If Pop and Phil are the best and they range from 12-17 wins what would an average coach's impact be? From no impact to maybe 5 games?
If I had significant time to create a statistical model normalizing effects of Roster Changes, Management Changes, Injuries, Schedule fluxuation, Changes in talent level of competition, Assistant Coaching impact, home and away scheduling impact....etc.....I would likely have to be retired. Unfortunately I run my own business, have wife, young kids and a house as well as a full time job.
My point is that just because someone creates a white paper regarding a statistical model - does not make it an absolute truth. For example, Note that there are counter arguments such as the Dean Oliver view presented in your OP. To lay the claim that there is an absolute range for coaching impact of 12-17 wins based on one analysis would not be the way I would go. But to each their own.
In order to better create a statistical control - I might take a situational micro view rather than looking at Coaching statistics as a whole. For example (an example for Knicks fans)........what was the impact of Woodson when he took over for D'antoni in 2011-12? D'antoni had a .429 winning percentage at 6 Games under .500 for half a year and Woodson had a .750 winning percentage of 12 games over .500 for the second half of the year. That's a net 18 game swing in effectiveness with the same roster for a coach who is not in the same universe as Pop & Phil. The point I'm making is that if the original theory presented in your OP was "fact".....then examples such as (Woodson) I just presented would likely not happen to that impact level.
All of those additional factors you bring up could be researched but IMO that doesn't actually prove that this particular approach is flawed. They reviewed 62 coaches over 30 years! All the coaches had to have at least 15 everyday players come to their team and 15 leave. Using the Wins Produced Metric to measure any change in the players performance. That's not a small sample size or limited set of data points. IMO it's a pretty smart way of approaching the question. It would in fact cover a lot of the things you're questioning.
Just so you know I purposefully added the points made by Oliver to add a countervailing view. Even Oliver had Phil at 12 win impact. The Woodson example is useless since it's such a small sample size that you can't take that seriously. I disagree with your point that a Woodson impact would not likely happen. The chances for outliers in short bursts is very possible but that's not the point.
I continue to disagree. However in the interest of moving forward - based on your opinion (based on the linked analysis) that coaches don't seem to matter all that much (unless you have a legendary type such as PJ or Pop).....what is your opinion about what the Knicks should do about the coaching situation?
My belief is that Coaches do matter and that Rambis in his career has proven to be a net negative coach and should be replaced for a shot at a net positive coach. This even works into the statistical analysis you present. If Rambis effect the Knicks by Minus 6 Games over the course of a season - and a New Coach comes in and effects the team as a Plus 6......that's a 12 point swing and could be the difference in Playoffs or not.
I actually do think we should hire a new coach, but my point is that it would likely only marginally improve, unless you can address the issues with the Roster. There's no doubt in my mind that if you added 2 top quality guards this team would win regardless of who the coach was. I think Rambis was 1-9 in close games or something like that. Those losses could easily have been wins with better talent and not a single thing changed in terms of coaching. Now it could equally be argued that a better coach could've managed the games better and pulled out some of those games. IMO that is harder to prove than the case of adding better players!
The reason I say this is that in terms of guards this team has one of the worst set of guards in terms of collective performance. Our best guard statistically is actually the guy most want gone! Jose performs at about league average and none of the other guards is above average. Adding at least 2 above average NBA guards would improve things tremendously.
Rk Player Age G MP PER TS% USG% OWS DWS WS WS/48 ▾
1 Jose Calderon 34 72 2024 12.3 .571 12.6 2.7 1.4 4.0 .096
2 Langston Galloway 24 82 2033 11.7 .490 15.6 1.4 1.8 3.3 .077
3 Sasha Vujacic 31 61 908 11.1 .503 13.0 0.3 0.8 1.1 .060
4 Arron Afflalo 30 71 2371 10.9 .531 17.9 1.8 0.9 2.7 .055
5 Jerian Grant 23 76 1265 11.2 .479 19.0 -0.2 1.0 0.8 .030
