Einstein

Well that’s definitely not true. It’s how pattern matching with Cover 3 works. It wouldn’t be possible if the boundary didn’t move to cover the flat, since the SS is moving to cover the boundaries original deep 3rd he abandoned on the man route. It becomes Mable with the SS. Saban defenses do this all the time. Agree to disagree I guess. I don’t see any banjo here at all.

You’re right, but here is my hypothesis on why that is happening. When the play first develops. He IS watching Cook and maintaining responsibility for the flat. If Allen throws it right now (pic below) to Cook, it’s the SS who is taking the flat tackle. But as the play developers and Diggs crosses the field, the SS pattern matches and takes Diggs, while the weak side third comes down to cover the flat. Which would basically be a Cover 3 Sky/Buzz with a pattern match, right?

Correct. The strong safety comes down to cover the flat. Which is exactly what we are seeing in the Tampa scheme. Except the strong side is flipped. To be honest it’s not even a question in my mind that this is some version of Cover 3 Sky. Not sue what you mean by they would all be wrong. They’re all where they mostly should be if they’re running Cover 3. .

I agree, this is not a banjo call. Im sticking with my original assessment that this is some subtype of Cover 3, with my best guess being Sky (just look at the down safety).

I disagree. This is what Allen is seeing (photo below). An open Kincaid with a large enough space between Allen and the free rusher to rifle a patented Josh Allen tomahawk missile, despite the tight alley. I truly believe if this same scenario was attempted 10 times in a row, Allen completes at least 8 of them. Winfield jumps in the air and barely gets one hand (maybe even half a hand) on the ball. It was a good play by Winfield, and Allen probably should have just ate the sack, but as mentioned before, he knows he can make it work more times than not.

Im pretty sure it was an attempt at Cover 3 Sky.

Wasn’t clearly downfield!? He was half way to Bahstan!

Not sure I agree here. Allen was plenty vertical enough to turf it or eat it. He chose to make the throw and it was tipped. Being a yard or two more vertical would not have changed his options. Agreed. Allen gets that throw off 9 out of 10 times.

My nana is off the boat from Italy and the first thing she said as she stepped foot on Ellis Island in 1910 was "Go Bills".

And the tape proves he is not. We are still running 10 and 11 personnel with Diggs running posts and fly's in the 4th quarter up 14 points. The offense never took their foot off the gas until the last drive when we ran 3 times to burn clock.

TNF must have really lowered the volume on the television broadcast because it felt the exact opposite watching it on TV. Very quiet.

Gosh, I honestly don’t know.

To the observer - yes.

No. And i'll ignore the insult. Primacy effect is a short-term effect (it is sequential of one event). A prior game would not be relevant. Psychology of Sport and Exercise 8 (2007) 477–489 Study "Primacy effects were observed regardless of the soccer experience of the participants and the judgement mode (end-of-sequence versus step-by-step).... This finding has a number of implications. First, it suggests that the way in which athletes’ abilities are judged may be biased by the order in which their performances are judged. The results indicate that it may be more beneficial for an athlete to attempt to start strongly than to start poorly, as this will produce a more favourable impression in perceivers." https://pubmed.ncbi.nlm.nih.gov/19437187/

It’s difficult to do, because he is a bad pass blocker. When the opponent brings more rushers, the RB has to stay in and protect. He struggles.

How many more points?

Yes. Perhaps a form of the cognitive bias known as Primacy Effect.

Likely not, because we then have to develop an error system that avoid flagging every single time an o-linemen grazes a d-linemens facemask while blocking (happens nearly every play). The system will need an incredible amount of redundancy.

Gotcha. I think that is a valid hypothesis.

We can't (with current technology). You can't simply "train" a model to know when a 3d object such as fingers grasp onto a facemask (rather than graze) in real time. You would need a 3D convolutional neural network that receives data from a LIDAR-type 3D camera producing volumes of data that can differentiate between individual fingers and the material of the facemask from multiple directions. It then has to compute this in real-time, billions of computations per second. I We are hoping that future advancements will enable the ability for us to do exactly what you are requesting. Maybe in another 10 years or so.

This was the first thing I thought of. But if you look at the tape, they were in attack mode until the very last drive. They were running 10 and 11-Personnel throughout. Aaron Quinn attacked the "foot off the gas" theory in a long Twitter thread today.

You feel like the offense was purposefully not scoring the last quarter and a half? The staff wanted to practice punting?

I made the argument last week that the offense played 'fine'. Points are points whether they are scored in the first half or final minutes. The offense scored 25 and left the field with the lead and under two minutes to play. So I figured lets compare the two. Week 7 vs Patriots Points: 25 Offensive EPA/a: .16 First downs: 24 Punts: 1 Week 8 vs Bucs Points: 24 Offensive EPA/a: .09 First downs: 25 Punts: 4 This weeks offense did have 88 more yards, and also involved more receiving options, though this effort resulted in 1 less point, a lower EPA/a and more punts than last week. We did have a large lead this week, which can contribute to taking the foot off the pedal. However, I didn't get the sense we were doing that until the very last drive. We were running 11 personnel with Josh in shotgun up until the second to last drive. PS, our no-huddle work was magnificent:

I read that Daquan Jones should be back for the playoffs. I believe it was by Dr. Chao.

I would like to see your prototype, if you're serious. This is already possible, but it is just tracking technology. Not what most would consider AI, though I guess one could consider any computer an artificial intelligence. These would be significantly more difficult. For example: How does the AI determine that the face mask was grabbed and pulled, rather than simply grazed? How does the AI know if the knee was down (this requires elevation, rather than simply distance) or if the referees whistle was blown prior to the ball coming out on fumble? That doesnt even get into latency, snow, error handling, etc. We are talking 1,000+ TFLOPS (with error).