Coach aggressiveness rating (1 Viewer)

[big windy]

I'm working on a method for describing coaches as "aggressive" or "conservative" that is going to feed some research that I want to do. Eventually, I'd like to attempt to answer questions such as, does playing extremely aggressively/conservatively when you are an underdog/favored or when you are losing/winning at the half result in a better than expected win percentage? Does it become detrimental and when? Does playing really aggressively lead to bigger wins but also bigger blowouts? Do some coaches seem to prefer small losses to big losses at the expense of improving their win probability?I wanted to ask you guys your thoughts on what makes a coach "aggressive" and what you think of my overall approach so far. I am ranking coaches by year on various attributes that act as proxies for aggressiveness, and then taking either a straight or weighted average of the various rankings. I'm looking only at 1st half results to come up with these aggressiveness attributes because I feel this better represents a coach's inherent philosophy. The data shows on average coaches have a very balanced approach until the start of the 3rd quarter, when run/pass mix begins to diverge very cleanly depending on the point margin. Some caveats: do all coaches have an overarching aggressive/conservative philosophy or is it sometimes a function of team ownership and roster? I can test this by looking at persistence of a coach's ranking as aggressive or conservative over time (and across teams in some instances). Do coaches behave differently if they have a star QB? This can also be controlled to an extent.For aggressiveness attributes, I've come up with deep pass rate, going for it on 4th down, blitz rate, and I'm on the fence with overall pass rate. Unfortunately, I can't find any play by play data or any aggregate data (broken about by half) that records blitzes. I'd be very grateful if anyone could help me find this data.My first attempt used pass%, deep passes as a percent of total plays, 4th down attempts as a percent of total 4th downs, and DB sacks as a percent of total sacks. I weight the highest and lowest ranking for each coach at about 50% of the middle ranks to reduce the effect of major outliers. Pass-happy offenses dominate these rankings, and that worries me because is an offense using a lot of short, low-risk passes actually that much more aggressive than a team with a strong running game, overall?Here are the top 10 2006-2011:2010_SEA_Pete Carroll2009_CHI_Lovie Smith2011_GB_Mike McCarthy2010_DAL_Wade Phillips2007_DET_Rod Marinelli2006_PHI_Andy Reid2009_SD_Norv Turner2007_NO_Sean Payton2009_NO_Sean Payton2010_ARI_Ken WhisenhuntAnd the bottom 10:2008_CLE_Romeo Crennel2008_WAS_Jim Zorn2011_IND_Jim Caldwell2006_JAC_Jack Del Rio2009_DAL_Wade Phillips2009_CAR_John Fox2011_HOU_Gary Kubiak2010_CAR_John Fox2011_MIN_Leslie Frazier2007_MIN_Brad ChildressI think a good portion of the top 10 make sense, and especially the bottom 10, when I think about aggressive/conservative coaches or team seasons.My second attempt took out overall pass rate, leaving only deep pass rate, 4th down attempts, and DB sacks. Remember that I'm using rates to control for teams that have more/fewer attempts because they are better/worse. This resulted in the following top 10:2008_STL_Scott Linehan2010_SEA_Pete Carroll2007_PIT_Mike Tomlin2009_CHI_Lovie Smith2011_GB_Mike McCarthy2007_STL_Scott Linehan2009_NO_Sean Payton2010_MIN_Leslie Frazier2010_ARI_Ken Whisenhunt2009_SD_Norv TurnerWade Phillips, Rod Marinelli, Andy Reid, and one of the Sean Paytons were replaced by Scott Linehan x2, Mike Tomlin, and Leslie Frazier.And bottom 10:2006_BAL_Brian Billick2011_IND_Jim Caldwell2010_DET_Jim Schwartz2011_SF_Jim Harbaugh2008_WAS_Jim Zorn2011_HOU_Gary Kubiak2007_MIN_Brad Childress2010_STL_Steve Spagnuolo2011_MIN_Leslie Frazier2009_DAL_Wade PhillipsLots of Jims here. Romeo Crennel, Jack Del Rio, and John Fox x2 were replaced by Brian Billick, Jim Schwartz, Jim Harbaugh, and Steve Spagnuolo.Leslie Frazier makes the bottom 10 and the top 10 in different years under the same rating method, which suggests that desperation or experience may be more of a driving force than a strict philosophy in some instances. I'm only showing the top and bottom of the rankings here, but there are many more instances where a coach can flip flop year to year.If you guys find this interesting, I'll post more results as I go, but I won't be finished with my initial work until the end of March. If there seems to be interest, I'll try and make periodic updates until then. Any insight or opinions would definitely be appreciated.

 

[grateful zed]

My link

 

[coolnerd]

Houston fans would hae placed Gary Kubiak in the bottom ten. where does he fall by your methodology

 

[big windy]

2011 Gary Kubiak made the bottom 10 when factoring in pass% and unfortunately I started this before the 2012 data was available or I'm sure 2012 Kubiak would be down there also.. His first few years in Houston he actually ranks pretty high when considering pass%. This appears to be an instance where a coach was playing aggressively until he had the right pieces in place to play the type of game he prefers.I was surprised Belichik didn't make the top 10 by either measure, but according to my rankings he's generally mid-high. This makes some sense if you assume coaches adjust their aggressiveness conditional on the talent of the team. The Pats don't need to be overly aggressive because they are favored to win most games. What we see is Belichik acting more optimally rather than aggressively. One of the theories I want to test is that coaches adjust their risk preferences given their probability of winning. If you are an underdog, increasing the volatility of play increases your win probability, but also increases your chance of a huge blowout. My gut says some coaches adjust aggressiveness given a Vegas point spread or half time deficit while others are afraid of big losses. Or, they are averse to playing more optimally in the 2nd half because they dont want to be lambasted for "running up the score" or whatever. Winning may be the most important factor in front office evaluation of coaches, but my hunch is that some coaches of bad teams would rather beat the spread and lose than have huge blowouts because the media, fans, and ownership don't always understand the relationship between volatility and win probability. On the other hand, I think some coaches would rather blow a lead in the 2nd half by being too conservative than risk blowing a lead by playing optimally and passing more/going for it on 4th and short. Given how much Belichik is vilified by the media and non-Pats fans, I would think this is the bigger effect.One thing I really like about my 2nd attempt at capturing coach risk preferences (aggressiveness) is seeing Scott Linehan at the top. Without getting into how good of a coach he was in general, we at least see that he took one of the worst teams in the league and tried to win games he had amost no chance of winning by throwing deep often, going for it on 4th down, and blitzing more often. He got his butt handed to him and got fired but at least he tried.Disclaimer: When I talk about coaches being too conservative or not playing optimally, I do mean based on what we know empirically. Not everything can be measured with stats. The unknowable "truth" is probably somewhere between what the numbers say and what coaches actually do.

 

[encaitar]

There's way too much that goes into it to use stats to determine whether a coach is aggressive or conservative. You already mentioned whether a team is winning or losing is going to change how they approach the game. However, you fail to mention personnel. You already talked about how Leslie Frazier goes from the top 10 in 2010 to the bottom 10 in 2011. What changed? 2010 Leslie Frazier had Brett Favre as QB and Sidney RIce as a deep threat receiver. 2011 Leslie Frazier had Donovan McNabb/Christian Ponder as a QB and no deep threat WRs to speak of. Clearly his best chance was to hand the ball to AP. Call that conservative if you want, but it was really the only play he had.

 

[big windy]

All great points and I'm not fooling myself into thinking I'll find the best possible measure. I tried to incorporate multiple attributes to measure aggressiveness so that no one attribute dominates the rankings. I think 4th down attempt rate outside of a team's own 20-40 is the most neutral measure. I really wish I had actual blitz data because sacks are outcomes, not attempts, and more reflective of defensive ability. That's why I'm using DB sack rate since sending your DBs is probably a riskier play than sending an LB.I realize my first post was way too long, but I did mention that a caveat is that coaches have to play to their roster to some extent. I used the early versus later Kubiak years in Houston as an example. But I don't think it's a stretch that owners bring in coaches with a specific philosophy and want him to build the team in his image. It will be interesting to look over the tenure of a coach with a team, and the same coach across teams.It's totally fine if coaches change year to year. I can still look at teams who won as underdogs and back that out to which teams won these games, and did more aggressive teams win more of these games or does it mostly look like noise.The Frazier leaning on ADP as his only option is a great point I was trying to make. The spread is set assuming Frazier plays to his team's strengths, which are lacking. If the odds makers knew he was going to play more aggressively, then they would (in theory) tighten the spread because they would be less confident in the outcome. Another good example is Jeff Fisher and the Rams. The Rams did pretty well ATS this year. These spreads were based on expected points for and against at a given level of volatility (standard deviation). In my opinion, Fisher's gutsy calls and trick plays were an attempt to increase volatility and therefore increase his probability of winning. I haven't tested this, but I'd like to know if Vegas began tightening spreads later in the year based on the unpredictability Fisher introduced into games.

 

[Bracie Smathers]

I think you would also need to look beyond the stats and also find some way to incorporate aggressive and/or controversial personnnel and coaching decisions as we saw last year with both of the Super Bowl teams.Baltimore replaced their OC late in the season when they were a playoff team.Unherd of, highly controversial at the time but ver-VERY aggressive coaching decision that worked out. Jim Caldwell is listed as one of the least aggressive coaches but John Harbaugh made very personal and emotional coaching change and it worked out.Baltimore also replaced rookie right tackle Kelechi Osemele and moved him inside at playoff time.Two huge and very aggressive coaching moves by John Harbaugh.John's bro, Jim Harbaugh, also made an aggressive coaching move when he replaced his starting quarterback, Alex Smith, when Smith got injured. When Alex got injured he either had the top QB passer rating in the league or was one of the top guys. The switch from Smith to Kaepernick meant a change in offensive philosophy and anytime a QB change is made its controversial but when a playoff caliber team replaces a league leader at QB passer rating with an unproven second-year guy its highly-HIGHLY controversial and can be viewed as an aggressive coaching move.So their would have to be some way to factor in coaching moves that are not noted in stats.Basically I think it might be hard if not impossible to log or take note of things like the above examples because many aggressive coaching moves don't pan-out and hit pay-dirt where we saw and noted that two coaches made highly aggressive coaching moves and made the Super Bowl in the same year.Those things are fresh in our minds right now but I think they were HUGE moves and the aggressiveness of the Harbaugh brothers should be noted by other NFL head coaches and front office types.

 

[GregR]

For deep passes are you using complertons that gained a lot of yards, or attempts that traveled far in the air including incompletions?

 

[Biabreakable]

Interesting stuff. I just wanted to add to encaitar's point about the Vikings and Frazier.In 2010 with Brett Favre he changed the play so often that there was even a point where Childress was talking about Benching Favre because he was not following the game plan. So I would consider that year to be how Favre calls plays on offense moreso than Childress or Frazier.Frazier is without question a conservative coach who wants to win with a good running game, defense and special teams. At the same time Frazier does talk about explosive plays (plays over 20 yards) and that explosive plays can come from the running game or the passing game. I think a explosive rushing play is over 15 yards rather than 20 to Frazier but would need to look that up again. That could be a criteria for testing aggressiveness I suppose? The explosive plays, but I am not sure that really tells us how aggressive/conservative the coach is. It may just be how well the team executed.

 

[big windy]

For deep passes are you using complertons that gained a lot of yards, or attempts that traveled far in the air including incompletions?

Attempts that traveled far in the air including incompletions. The pbp data I'm using has a "location" field that charts passes as short/medium/deep and left/right/center. This is great because passes that went for a lot of yards don't tell us much when there's a Percy Harvin on the field.

 

[big windy]

Interesting stuff. I just wanted to add to encaitar's point about the Vikings and Frazier.In 2010 with Brett Favre he changed the play so often that there was even a point where Childress was talking about Benching Favre because he was not following the game plan. So I would consider that year to be how Favre calls plays on offense moreso than Childress or Frazier.Frazier is without question a conservative coach who wants to win with a good running game, defense and special teams. At the same time Frazier does talk about explosive plays (plays over 20 yards) and that explosive plays can come from the running game or the passing game. I think a explosive rushing play is over 15 yards rather than 20 to Frazier but would need to look that up again. That could be a criteria for testing aggressiveness I suppose? The explosive plays, but I am not sure that really tells us how aggressive/conservative the coach is. It may just be how well the team executed.

Yeah that seems like a really good point about QB audibles. Luckily, at least I think, there aren't enough veteran gunslingers with the attitude and fan/front office support to influence the data too badly. And on a year-to-year basis, Childress always ranks near the bottom, even in 2010, so it appears the other attributes are evening things out. From 2006 to 2010, Childress ranked 22nd, 30th, 30th, 25th, and 33rd (due to firings there were more than 32 coaches in a given season). Frazier ranked 13th and 27th in 2010 and 2011. It seems like Childress was so conservative (most in the league) that Frazier felt like he needed to show a significant change in style to get extended as HC in the 2011. This is actually kinda cool because most of the 1 and done coaches in the sample period are interim stop gaps.A tendency to want to run explosive plays would be a great thing to know, but like you mentioned, it's too difficult to separate those outcomes from play calling versus execution.

 

[GregR]

For deep passes are you using complertons that gained a lot of yards, or attempts that traveled far in the air including incompletions?

Attempts that traveled far in the air including incompletions. The pbp data I'm using has a "location" field that charts passes as short/medium/deep and left/right/center. This is great because passes that went for a lot of yards don't tell us much when there's a Percy Harvin on the field.Sounds good.Was giving this some thought... if you don't think you're getting the results that you want... you could consider combining down, distance, and play type (run/short pass/medium pass/long pass) to give every play some sort of an aggressiveness rating for offensive plays.Just using run/pass rate won't differentiate between calling a conservative run on 3rd and 15 versus a "normal" run on 3rd and 1. A long pass attempt on 1st and 10 is more aggressive than a long pass attempt on 3rd and 15. You'd need to do some thinking on what would be equivalent aggression. Is a 1st and 10 run the same aggressiveness as a 2nd and 6 run, and as 2nd and 10 short pass? Once you decide what should be about the same then figure out how to incorporate all 3 factors into calculating a final number that when compared ranks the plays in the right order.

 

[JAA]

I'm working on a method for describing coaches as "aggressive" or "conservative" that is going to feed some research that I want to do. Eventually, I'd like to attempt to answer questions such as, does playing extremely aggressively/conservatively when you are an underdog/favored or when you are losing/winning at the half result in a better than expected win percentage? Does it become detrimental and when? Does playing really aggressively lead to bigger wins but also bigger blowouts? Do some coaches seem to prefer small losses to big losses at the expense of improving their win probability?I wanted to ask you guys your thoughts on what makes a coach "aggressive" and what you think of my overall approach so far. I am ranking coaches by year on various attributes that act as proxies for aggressiveness, and then taking either a straight or weighted average of the various rankings. I'm looking only at 1st half results to come up with these aggressiveness attributes because I feel this better represents a coach's inherent philosophy. The data shows on average coaches have a very balanced approach until the start of the 3rd quarter, when run/pass mix begins to diverge very cleanly depending on the point margin. Some caveats: do all coaches have an overarching aggressive/conservative philosophy or is it sometimes a function of team ownership and roster? I can test this by looking at persistence of a coach's ranking as aggressive or conservative over time (and across teams in some instances). Do coaches behave differently if they have a star QB? This can also be controlled to an extent.For aggressiveness attributes, I've come up with deep pass rate, going for it on 4th down, blitz rate, and I'm on the fence with overall pass rate. Unfortunately, I can't find any play by play data or any aggregate data (broken about by half) that records blitzes. I'd be very grateful if anyone could help me find this data.My first attempt used pass%, deep passes as a percent of total plays, 4th down attempts as a percent of total 4th downs, and DB sacks as a percent of total sacks. I weight the highest and lowest ranking for each coach at about 50% of the middle ranks to reduce the effect of major outliers. Pass-happy offenses dominate these rankings, and that worries me because is an offense using a lot of short, low-risk passes actually that much more aggressive than a team with a strong running game, overall?Here are the top 10 2006-2011:2010_SEA_Pete Carroll2009_CHI_Lovie Smith2011_GB_Mike McCarthy2010_DAL_Wade Phillips2007_DET_Rod Marinelli2006_PHI_Andy Reid2009_SD_Norv Turner2007_NO_Sean Payton2009_NO_Sean Payton2010_ARI_Ken WhisenhuntAnd the bottom 10:2008_CLE_Romeo Crennel2008_WAS_Jim Zorn2011_IND_Jim Caldwell2006_JAC_Jack Del Rio2009_DAL_Wade Phillips2009_CAR_John Fox2011_HOU_Gary Kubiak2010_CAR_John Fox2011_MIN_Leslie Frazier2007_MIN_Brad ChildressI think a good portion of the top 10 make sense, and especially the bottom 10, when I think about aggressive/conservative coaches or team seasons.My second attempt took out overall pass rate, leaving only deep pass rate, 4th down attempts, and DB sacks. Remember that I'm using rates to control for teams that have more/fewer attempts because they are better/worse. This resulted in the following top 10:2008_STL_Scott Linehan2010_SEA_Pete Carroll2007_PIT_Mike Tomlin2009_CHI_Lovie Smith2011_GB_Mike McCarthy2007_STL_Scott Linehan2009_NO_Sean Payton2010_MIN_Leslie Frazier2010_ARI_Ken Whisenhunt2009_SD_Norv TurnerWade Phillips, Rod Marinelli, Andy Reid, and one of the Sean Paytons were replaced by Scott Linehan x2, Mike Tomlin, and Leslie Frazier.And bottom 10:2006_BAL_Brian Billick2011_IND_Jim Caldwell2010_DET_Jim Schwartz2011_SF_Jim Harbaugh2008_WAS_Jim Zorn2011_HOU_Gary Kubiak2007_MIN_Brad Childress2010_STL_Steve Spagnuolo2011_MIN_Leslie Frazier2009_DAL_Wade PhillipsLots of Jims here. Romeo Crennel, Jack Del Rio, and John Fox x2 were replaced by Brian Billick, Jim Schwartz, Jim Harbaugh, and Steve Spagnuolo.Leslie Frazier makes the bottom 10 and the top 10 in different years under the same rating method, which suggests that desperation or experience may be more of a driving force than a strict philosophy in some instances. I'm only showing the top and bottom of the rankings here, but there are many more instances where a coach can flip flop year to year.If you guys find this interesting, I'll post more results as I go, but I won't be finished with my initial work until the end of March. If there seems to be interest, I'll try and make periodic updates until then. Any insight or opinions would definitely be appreciated.

Really interesting no Billicheck in here

 

[FF Ninja]

Kubiak's spot looks about right. He might've set some records in 2012.

 

[big windy]

Really interesting no Billicheck in here

Yeah, I was surprised by that as well. But then I thought about it some more...is he overly risk-seeking or is he just playing optimally? Since the Pats are already a really good team, they have one of the highest expected points ratios (Points Scored/Points Allowed). All else equal, the Pats want to keep the volatility of the game minimized. In other words, if the Pats are expected to win based on skill alone, they want to minimize how much chance/luck plays in to the outcome. If you look at the guy at the top of the second set of lists, Scott Linehan, he was playing with the least talented team in the league, and therefore it makes sense that he wanted to increase the level that luck/chance played in the game, because even though this style of play would lead to bigger blowouts when the lost, it also increased the Rams' chances of winning games they weren't supposed to win.