You don't need everything else to remain the same. In the simplest approach you do multivariate regression, which assumes there is random variation when it estimates the effect of different factors. The more complex black box approach which I bet some like Amazon will do is to use machine learning. The only validation of that is predictive power, but if it works then great. It's definitely something that can be done. It's just not done because it's not an important area of research for any serious scientist. I mean seriously.. who cares from a societal point of view how much a QB affects win% lol.
Given the fact that the NFL is a multi-billion dollar business, and player salaries are hundreds of millions of dollars, I'd say there are at least a few dozen billionaires that care. I can tell you that the current NFL believes QBs have the largest impact.
And analytics is increasing its influence in the NFL. But they don't hire the best people from a data analytics perspective that's for sure, not even close. Even sabermetrics in baseball was developed from outside MLB. The establishment actually fought it until it became undeniable that it helped increase win% after which they hired people competent in statistics but hardly anyone who stood out in the world of statistics. There's no important mathematician or statistician who made key discoveries looking at sports data. So I think the more accurate statement is that while there is clearly a monetary incentive for owners and coaches to improve analytics, they don't act like it matters much when you look at where the money goes. They don't fund research the way they should if they really wanted the competitive advantage you'd likely see by funding that research.
I think Ernie Adams was the statistician that helped the Pats all those years. https://www.espn.com/espn/eticket/story?page=adams&redirected=true "Adams and Belichick were known to have an interest in mathematical analyses of football; Adams once contacted Rutgers University statistics professor Harold Sackrowitz, asking him to evaluate the Patriots' two-point conversion chart following a study by Sackrowitz on the decision.[2][3] Adams also presented Belichick with a study concluding teams punt too often on fourth down." https://en.wikipedia.org/wiki/Ernie_Adams_(American_football)
As I read these posts, I can’t help but shake my head, how so many “absolute” opinions are made only on partial variables. You all do recall football is a TEAM game played by ELEVEN MEN, each with different RESPONSIBILITIES, all requiring PRECISE EXECUTION in order to effect a desired outcome, right? Coaches calling the correct play at the right time Offensive linemen executing precise blocks Running backs hitting holes or cutting outside times with the designed blocking for said okay Receivers running precise routes, getting off coverages AND catching throws within the grasp. Quarterbacks pre-snap readings audibles as needed, precision in handling the ball off to running backs, effectively using the time provided in the pocket to accurately put the ball into the right receiver’s hands. None of these stats can be accurately absolute in a vacuum. Passer rating for example is a mathematical evaluation on 2 of 11 variables, the quarterback and the receiver. He either caught the ball or he didn’t. The quarterback either threw a completed pass, an in competition or an interception. Passer rating doesn’t take into account whether or not the right play was called, whether the quarterback had adequate blocking to allow a play to develop or whether a receiver failed to get his route or get open. Tannehill this…Tua that. Well, how about the other 10 guys on the field…and the head coach on the sideline? If football was a theoretical team sport like basketball or baseball, all a team would have to do is out the ball in the hands of their LeBron James or Greg Maddox and ride that player to victory. Sadly, that’s not how this game works.
Interesting. Broadly speaking I'd create 4 categories of "statisticians" for purposes of this discussion. 1. Knows some statistical analysis but not sufficiently proficient to apply the correct statistical analysis to all relevant data. 2. Well enough versed in statistical analysis to apply a justifiable method to the data and obtain a relevant result 3. Ability to develop new statistical methodology considered at least somewhat important in the field. 4. Has developed new and widely used statistical methodology that is considered historically important I think Adams is in the #1 category if he had to ask this professor to look at something as simple as two-point conversions. The professor looks like he's probably in #3. His research is mostly not related to sports, which means he might be doing this as a consultant more than anything else. Question is whether there is anyone in #3 who primarily does football-related analysis. Not sure. Regardless, there is no one in #4 who made their name on any kind of sports statistics which really emphasizes how little value is placed on it.
If we didn't use stats because they're not perfect we'd just have opinions, many of them unsupported without any objective evidence. The reason stats are becoming more and more influential in all aspects of the NFL, from personnel decisions to play calling, is precisely because impoverished but objective data analyzed with mathematics is often more valuable than rich data analyzed poorly by humans using their intuition and experience. Stats increase win%. NFL teams will tell you that, which is why we shouldn't treat them the way you're suggesting (unless you can prove your ability to predict win% is way better than statistics, but it's not — very hard to beat these models).
If we know Ernie and BB did this, who knows what they really did and with whom else they contacted? I'm sure there's some #4('s) out there who love football and would happily develop something useful to winning...especially if Kraft and company were willing to pay them...and if they did they'd certainly make him/her/them sign an NDA to not release what they developed. Point being...we know the Pats cheated. We know Ernie Adams was akin to the ghost in the machine and we never really heard much about him because they tried to keep him a secret. We can deduce that Adams was a HUGE factor in their winning. Is it crazy to think that they hired someone to develop a process and Ernie was good enough to run with it? I mean, the 20 year run they had was so incredible. Almost impossible without some outside influences.
I have no problem with using stats but as my tag line says - Film first, numbers second. If the numbers don’t match what is seen on film, something is likely wrong with the numbers.
Well, you can look at the Earth from your point of view and your eyes will tell you it's flat. I'm not saying that watching film isn't a great way to evaluate a player. However, properly created stats will always beat a human eye. The problem is, it's hard to get good stats...which is what cbrad has been saying. Although, some of these imperfect stats will still do better at predicting wins/losses than a person watching film. If watching film was so great there wouldn't be so many 1st round busts in the NFL, for example.
The last 2 sentences are the key right? What exactly is it that led to that. As much as I mention Brady he wasn't the only elite QB on a good team. The only comparable dynasty was the 49ers but that was easier without the salary cap. My guess: they were just a tad ahead of the curve in obtaining data (legally that is.. that technical illegality of filming from the wrong place doesn't mean much in the grand scheme of things) and analyzing the data. They might be the best example of the use of analytics we don't yet know the full story of, but I doubt it was anything fancy. Sabermetrics made a huge mark just by comparing player stats to those of replacement level players. You don't need much brains to do that, but the effect can be huge if you're doing it and others aren't. Hopefully we hear the full story someday. I bet it was just a bunch of smaller things that accumulated in significance rather than any brilliant insight here or there.
The filming from the wrong spot was a cover. The real films that showed the truth were smashed with hammers and stepped on. That wouldn't have happened if it was only taping from the wrong area. Rams players, and other coaches and players, have openly talked about the Pats cheating ways. They've probably been hushed so as not to reveal what they really know. Being ahead of the curve for a few seasons is one thing, but being ahead for 20 years is something entirely different. I'll never be able to give BB and Brady full accolades because their dynasty is tainted. And I fear we may never hear the entire story unless someone confesses on their death bed.
https://www.businessinsider.com/espn-report-patriots-spygate-scandal-2015-9 Other methods of cheating reportedly include: Sending low-level Patriots employees to sneak into the visiting locker room during pregame warm-ups and steal the play sheet. As Van Natta and Wickersham note, "The practice became so notorious that some coaches put out fake play sheets for the Patriots to swipe." Sending employees through the visiting team's hotel to try to swipe playbooks and scouting reports. "Scrambling and jamming" opponent radio headsets to interfere with opponent communication. Roger Goodell, then only 18 months into his role as NFL commissioner, fined Belichick $500,000, the Patriots $250,000, and docked the team a first-round pick before investigators ever even went to Foxborough to look for the tapes. There, investigators found "a room accessible only to Belichick and a few others" with a "library of scouting material containing videotapes of opponents' signals, with detailed notes matching signals to plays for many teams going back seven seasons." However, as Van Natta and Wickersham report, "almost as quickly as the tapes and notes were found, they were destroyed, on Goodell's orders: League executives stomped the tapes into pieces and shredded the papers inside a Gillette Stadium conference room."
It’s a small sample size, but look at this highlight reel of Chad Henne. He was FAR from a celebrated quarterback in Miami, but if you look closely at each and every play, incompletions were not his fault and yet…
That's interesting, but how do you explain this no longer being effective once Adams left? When someone is in place like that for 20 years there are more than enough people who worked with him that can continue that after he leaves. Also, one would expect people in other organizations to emulate this. Remember, Jimmy Johnson admitted to doing exactly the same thing NE was accused of and called stealing signals a "common practice" that many other teams did. So something still doesn't fit here.
No question Henne was in a bad situation. The problem with Henne is he never improved the offense of any team he went to. Miami was bottom 1/3 in points scored prior to Henne, and over 2 years of Henne starting remained that way. Same with Jacksonville which was one of the worst teams on offense by points scored prior to Henne and only got worse with Henne (they ended up the worst team on offense with Henne his 2nd year starting). Henne never elevated the team. His below average QB reputation is deserved.
I don't think there were many people involved and the method left with him. He had it made/perfected it/kept it secret. Ernie and BB didn't care what happened to their method after they left. They took it with them. Jimmy Johnson, and others, downplayed the cheating, but who told them to do that? And there's a difference between happening upon a teams signals and creating a systematic and long running method of stealing them. Not all of the Pats success was due to cheating, as I've written I think stat compiling, or a method of stat compiling also helped, but add the cheating and that put them over the top. And again, this is what we know about...I can only imagine what never came out. *Edit: There was actually a phone labeled "Ernie" on the Pats sidelines. Only BB used it. Those two had a system and those two were the only ones who knew the inner workings. Sure, they had people stealing plays, etc...but those people had no idea what to do with them.
One other problem that once again suggests Brady was the more important factor. Adams left after the 2020 season while Brady left after the 2019 season. So Belichick + Adams had the entire system in place after Brady left. They went from 7th in points scored to 27th in one year, from 12-4 to 7-9. I do agree with you that something needs to be explained given that NE's success was only replicated one other time by SF in NFL history and SF didn't have to deal with the salary cap. But it still coincides more with Brady leaving than Adams.
In 2021 the Pats had the 6th ranked offense with Mac Jones and made the playoffs. Maybe Newton was just too horrible? Maybe Bill was able to keep doing Ernie's work for a bit? All I know is that Ernie Adams was extremely important there. Why? Most players had no idea who he even was. And the 49ers had that great 1981 to 1994 run, but like you said no salary cap, and they had some all time greats on defense, offense, and coaching. The Pats fielded guys who played great in NE, but typically looked liked trash elsewhere.
Also don't forget Adams was there in 2000 with Bledsoe and they went 5-11, next year Brady comes in they go 11-5 and win the SB. So Belichick + Adams was there one year before Brady and one year after Brady and in both cases you see a massive change. Belichick + Adams was also there with Cleveland and you don't see anything special. Even including all the post-Brady years you get that 5 win difference between Belichick + Brady vs Belichick. So I think the evidence shows the most important factor here is Brady. That changes the problem a bit. Yes, something needs to be explained, but the effect of Adams doesn't have to be that great anymore. You have a great defensive mind + a historically great QB and all you need is a little additional extra (Adams) to be more consistent at winning which could be explained by the accumulation of many smaller things, including just being smarter at analyzing legally obtained data on the opponent. Hard to say, but the problem isn't as severe once you see how it correlates so well with Brady. Also, it's worth noting that the only other time we saw a similar dynasty was with some of the greatest QBs ever in Montana and Young. The QB part of the equation is one commonality between the two dynasties. Walsh prior to Montana was a losing coach and the first 2 years after Young left led to losing seasons for Mariucci. Huge effect of QB in all these cases.
I get what you’re saying but look at the situations he was in his entire career. Sparano in Miami Mularky Bradly and Marone in Jacksonville By the time he got to a winning coach with Reid in Kansas City, he was already destroyed and washed up. I make the same comparisons with Tannehill. By the time he got to a good winning situation in Tennessee, his beat years were already behind him. Im so THANKFUL Tua only had to endure two seasons under Flores. He could very well have ended up in Henne/Tannehill territory had he had to endure Flores for longer than he did.
I could tell them what needs to be done for sure. Have too many things on my plate to do it myself. But the basic strategy would be to hire a team that understands football well enough to label video with new football-relevant features — it's essential to develop new statistics, not just analyze commonly used stats — and then have someone with programming skills (a CS guy specifically) test out different models, from simple linear models to different machine learning models. You then iterate by getting feedback from coaches on what's important and what isn't, relabel video with an updated set of new football-relevant features, and fit models to data. Guaranteed that gives you better analytics than what I've seen thus far. That labeling btw is very time consuming. Need a whole team doing that.
I don't think he's forgotten our arguments. But thanks for making sure. He's referencing that we were on the same side of the Tannehill debate. Also you misrepresent my position all the time. I've said that I think that Tua does now with elite weapons than others soe, but I think his play drops off more with lesser players than other QBs. It's fine, as long as we are able to keep those type of weapons for him. I've said many times that I believe Tua will have a good career barring injury.
lol no resnor. You are literally famous around here for constantly misrepresenting your own position, saying one thing then acting like you never said that even a few posts later. Even here, you're misrepresenting your own view by acting like what I said is not true. To refresh your memory: even after you finally relented and admitted Tua did more with his weapons than an average QB you never went back on your claim that without those weapons he's not much more than an average QB. That's precisely what I wrote in the post you falsely claimed misrepresented your position. If you want to restart the process of me quoting posts from you to literally disprove your numerous misrepresentations of your own position I can do that. Up to you.
I think it took a little time to get their cheating perfected. You can't steal signals before there are games played. My belief is Bledsoe could have did what Brady did, but he got hurt. Bledsoe was as good or better than Brady was over Brady's first couple of seasons. Remember, Bledsoe had been a regular in the playoffs, and even went to the SB, with worse defenses and coaches than Brady.
Still doesn't explain why it didn't work in Cleveland. I mean, did Belichick + Adams only try to cheat in NE? As far as Bledsoe, he had a mid-70's rating in 2000 (and the 2 games prior to injury in 2001) while Brady was mid-80's. Over the next 2 years in Buffalo Bledsoe averaged ~80 while Brady was mid-80's. And Bledsoe's overall playoff rating for 6 games is an abysmal 54.9. Brady > Bledsoe even in Brady's early years. So it's still the case that the best predictor for Belichick's success is Brady. I agree Adams was really important, and maybe without him they don't win as many SBs, but evidence suggests the primary difference was Brady.
Maybe they tried, maybe they didn’t. Maybe they didn’t have their eureka moment until 2000? Who knows. We definitely know that the Pats cheated. And you don’t risk doing what they did unless you think it’s going to help win games. On top of that, there’s so much more evidence out there that what they did went far beyond filming from the wrong area. And the NFL covered it up. Why would they cover it up? Probably due to the fact that the Federal Government is heavily involved with the NFL due to their anti-trust exemption and not-for-profit status. If those are revoked who knows what happens to the NFL. In Bledsoe’s previous two seasons he had an 87 and 81 PR. Brady finished the 2001 season with an 86. In 2001, Brady had a few outlier games, but in the majority of his starts he had PR’s in the 60’s and 70’s. 9 of his 15 games were 79 and below. 6 of which were 63 and below. He had 3 games over 100. Against the Colts and their 31st ranked defense, the Saints and their 27th ranked defense, and Atlanta with their 24th ranked defense. He also couldn’t beat out Bledsoe in the 2000 or 2001 preseasons. Brady also had 1 TD, 1 INT, and averaged less than 200 yards passing per game (with 115 yards passing in the AFCCG and 145 yards in the Super Bowl).
True, but those stats aren’t really enough and they are compiled against inferior competition. And most scouts and evaluators are still typically of the “eye test” variety. That’s why I specifically mentioned it’s hard to get good stats. Again, the “eye test” has its place and watching film to evaluate a player has its merits. Trusting your eyes and disregarding stats is going to lead to the wrong conclusion, however.
OTOH, ignoring the film and trusting the stats is just as bad. Both are needed, but IMO film is more revealing because you can see what actually happened. Here is a test using stats: QB stats for a game: 10 of 20 for 120 yards 1 TD, 2 ints How much is the QBs fault? You have no idea.
Except for not beating out Bledsoe in the preseason there's no question Brady > Bledsoe in those early years. And the fact coaches thought he wasn't as good as Bledsoe from preseason just demonstrates how wrong the coaches were. That's not an indictment on Brady. It's an indictment on Belichick. Brady actually turned out to be better once they played him. Let's make the comparison simple and look at passer rating from 2000 through successive years. Bledsoe PR 2000-2001: 77.1 2000-2002: 81.6 2000-2003: 79.2 2000-2004: 78.6 2000-2005: 79.6 Brady PR 2000-2001: 86.1 2000-2002: 85.9 2000-2003: 85.9 2000-2004: 87.5 2000-2005: 88.5 Really no argument when you see that. As far as playoffs, Bledsoe didn't really play in the playoffs during that time so there's no direct comparison, but as stated before his career playoff rating is an abysmal 54.9 with 6 games started. Brady's lowest was 77.3 in 2001 and only rose from there. So no question Brady > Bledsoe in Brady's early years.
Just those stats aren’t good enough. However, if we see those same stats over and over we can be certain it’s his fault. Again, looking at film is needed, but ignoring stats and using just your eyes (as you initially wrote) will lead to the wrong conclusions.
It's a sample size question. With small sample size you go with eyes over stats, with large sample size you go with stats over eyes. At least if you look at the predictive power of statistical models you see evidence of precisely that. Only a small percentage of people can consistently do better than the best statistical models with large sample size in predicting outcomes.
You’re using Bledsoe’s stats without BB and Ernie. Apples to oranges. The cheating was a thing, no? There’s really no reason to believe that Bledsoe, if he had stayed healthy, wouldn’t have won a SB in 2001. He was more experienced, had already been in a SB, and was more athletically gifted than Brady.
I think we've established that the cheating didn't affect things too much even if I agree with you that Adams probably did some unknown behind-the-scenes stuff that helped NE gain a competitive edge. The single best predictor of NE's regular season, playoff and SB success is Brady, not Adams. And for an apples-to-apples comparison you have that first statistic of 77.1 vs. 86.1 for passer rating from 2000-2001. Either way Brady > Bledsoe in Brady's early years in that apples-to-apples comparison. Oh, and the reason to not think Bledsoe would have won a SB is his abysmal 54.9 career playoff rating. That's with 252 passing attempts. It's not to be dismissed.
I don’t agree that we established that regarding the cheating. We may not know how much it helped, but we know they did it for around a decade before getting caught. We know the NFL covered up what really happened. We don’t know what else they did after, but we know many NFL execs, teams, players, and coaches believed the Pats were cheaters before and after they were caught. And breaking down Brady’s 86 season we see a few outliers against 3 of the worst defenses in the league. Otherwise his play was around that of Bledsoe’s.
