East Coast Offense: When You're Wrong

This article is part of our East Coast Offense series.

When You're Wrong

This seemingly innocuous statement was tweeted into my feed, and it got me thinking:

On the one hand, it seems trivial -- a statement of probability about something happening is not invalidated just because the improbable happens. He said it was less than three percent, not zero percent, after all.  

But what if I threw four sixes in a row? That's 1 in 1,296 through dumb luck. Still, it could happen, and the less-than-one-tenth-of-one-percent statement would still be true. How about 10 sixes? Six to the 10th power is 60,476,166. You still cool with your initial estimate if I get 10 straight sixes? What about 20? Now we're talking 1 in 3.65 quadrillion. Maybe somewhere between four and 20 sixes, we need to come up with a better explanation than dumb luck, such as weighted dice. In other words, there is a point where the improbable does in fact make your priors dead wrong. 

Of course, one could reply, when he was talking about dice in the hypothetical, surely he meant unweighted ones. I'll go ahead and assume he meant ideal wind conditions, a dice-thrower without any special skills to make one number show up more than others, etc. Let's sterilize the hypothetical, so to speak, to make each die truly one in six to show a six. In that case, of course, he is right, but the statement is meaningless, and there is no need to roll at all. We know in advance the precise probabilities because it's defined in the hypothetical itself. The tweet therefore begs the question. He is not wrong about the odds based on the improbable result because the premise of the hypothetical is that it is not possible for him to be wrong. 

That's well and good, but when you move away not only from virtual, perfect dice rolled in perfect conditions to the non-sterile real world of imperfect dice and farther still into the messy world of human beings playing a sport, you don't have foreknowledge that your model can't be falsified. Each real-life improbable event chips away at the likelihood your priors were in fact correct, and there's a legitimate argument that the "dice" so to speak were weighted. 

The difficult part is deciding how many "unlucky" results it takes before you realize your model was flat wrong. The idea contained in the tweet is a very special case that can only exist in a hypothetical and never in the real world, one where the model is true before the first flip, and reality is irrelevant to its measure. This might seem like a semantic or technical point, but, in my opinion, it's not. Those making probability-based models often analogize them to dice-rolling, not seeing the categorical difference that occurs when you make the jump into the real world. 

In related news I'm starting to think fading Derrick Henry based on his massive 2019-2020 workload was a mistake. 

Week 7 Sporcle

Apropos of Derrick Henry scoring three more rushing TDs Monday night, can you name every NFL player who scored 25 or more TDs from 2018-2021?

Guessing The Lines

I don't know what I was thinking with Broncos-Browns, but I suppose I'll go with it. Also the Dolphins strike me as the right pivot (though it's less than ideal to take a team back from London against another one off a bye). I also like the Saints in Seattle. Of course, I reserve the right to change my mind in Beating The Book.

Week 6 Observations

• Josh Allen (45 passes, nine runs) was worth whatever pick you spent to acquire him. The Bills are so pass heavy and also design runs for him.
• Stefon Diggs (11-9-89-1) had his best fantasy output of the year. It will be hard to keep up with last year's pace, but he's getting his looks. Cole Beasley (9-7-88-1) and Emmanuel Sanders (8-5-91-0) get theirs more often than not too.
• Ryan Tannehill was under pressure for much of the night, but made some good throws on key third downs. Better days are ahead with A.J. Brown back.
• Speaking of whom Brown (9-7-91-0) came back from an illness and had a big second half. Jones (5-3-59-0) had a more modest role.
• Derrick Henry (20-143-3, 3-2-15-0) will go down as one of the league's all-time great running backs. He had 784 carries from 2019-20 including the playoffs and seems none the worse for the wear.
• I'm glad i watched the Sunday night punt-fest via the 40-minute edited version.
• The Steelers cannot run-block, and Ben Roethlisberger doesn't have it anymore. Diontae Johnson (13-9-71-0) will rival Davante Adams for usage, though, and Najee Harris (24-81-0, 7-6-46-1) is basically peak Le'Veon Bell if Bell had a bad offensive line.
• Alex Collins (20-101-1) is no worse than Chris Carson. One cannot say the same thing about Geno Smith/Russell Wilson.
• I probably didn't downgrade DK Metcalf enough, and Tyler Lockett is borderline unstartable with Smith.
• Pats +3.5 was a tough beat, but I had a bad feeling when they made the two-point conversion after Kendrick Bourne's TD to go up three. Overtime not only distorts ATS results, but CeeDee Lamb's winning TD no doubt turned a ton of DFS and fantasy matchups too.
• Mac Jones showed poise after giving up the pick six to throw a perfect ball to Bourne for the go-ahead TD.
• The Raiders +3.5 – think the line moved up closer to game time – were always the right side. When it feels most wrong, narrative-wise, it's usually right.
• I thought the Browns would show up, but the Cardinals defense is better than I thought, and Cleveland had a ton of injuries.
• DeAndre Hopkins (4-3-55-2) got his thanks to the TDs, but I did not foresee him being just one of the guys in a receiving corps where A.J. Green (6-5-79-1) can easily lead the team in yardage.
• Baker Mayfield had one of the more egregious fumbles you'll see, then stayed down on the turf, apparently re-injuring his shoulder. He returned shortly thereafter, but not before prompting this tweet:

• I naively took the Chargers and the points, but this was the converse of the above principle, namely, when it feels too easy, it's probably wrong. Amazed at how well the Ravens defense played, though.
• Justin Herbert's terrible day also brought down Lamar Jackson whose heroics weren't needed.
• With Le'Veon Bell, Devonta Freeman, Latavius Murray and Sammy Watkins, the Ravens have a ton of 2015 fantasy value.
• Sam Darnold is showing Adam Gase wasn't the only problem.
• The Vikings didn't exactly ease Dalvin Cook (29-140-1, 2-2-3-0) back from his ankle injury.
• The Football Team is a doormat, at least with Taylor Heinicke.
• The bad Danny Dimes showed up for the first time this year with three picks and a lost fumble.
• Kadarius Toney (3-3-36-0) looked like he was in for a monster day before aggravating his ankle on the opening drive.
• Cooper Kupp (12-9-130-2) seems to get open on every play.
• Davis Mills and the Texans can effectively dink and dunk, using all four downs, but even after they got behind three scores, that was their only mode. They were chewing up clock late in the third quarter as if they had a lead.
• Carson Wentz connected on a couple deep passes early, but it seemed like he got greedy, holding the ball too long and taking sacks instead of dumping it off to open checkdowns.
• Jonathan Taylor (14-145-2, 2-1-13-0) was ignored in the first half, but broke the game open in the second. He's still a top five-ish back.
• Michael Pittman (3-2-35-0) was the victim of game flow, but T.Y. Hilton's return was also a factor.
• The Lions were a strong 0-5, but now they're just an above-average 0-6.
• After running for a late TD, Aaron Rodgers said to Bears fans, "I own you." But that was less funny than this response from a Bears fan:

• Aaron Jones (13-76-0, 4-4-34-1) had a split with AJ Dillion (11-59-0) more in line with my preseason expectations.
• Khalil Herbert (19-97-1, 3-2-15-0), ran well, but he dropped a pass, and Damien Williams and eventually David Montgomery are coming back.
• The game-tying 54-yard kick by Mike Wright was an obvious shank until it hooked back in, a miracle on par with Moses parting the Red Sea.
• Tom Brady looked sharp early, but the Eagles smartly took away the outside deep throws and forced him to hand off and dink and dunk. Game flow was also a factor as Brady was content to run clock and settle for longer drives, but almost all of his damage was done in the first half.
• Leonard Fournette  (22-81-2, 6-6-46) would go in the second round were we to re-draft for the rest of the season. He's the clear starter, looked smooth as a receiver and seems like the guy from last year's playoffs with his tackle breaking and vision. I made the comp to Marshawn Lynch, an early first-round prospect on whom his initial team soured, only to resurrect his career during the playoffs, and so far it holds.
• Antonio Brown (13-9-93-1) was Brady's favorite target. Mike Evans (4-3-27-0) and Chris Godwin (5-5-43-0) will get theirs in this offense, but it'll be almost random as to which one blows up in which game.
• O.J. Howard (7-6-49-1) did a decent Gronk impression. If he ever gets going, the riches will be beyond embarrassing.
• Jalen Hurts (4.4 YPA, 115 passing yards, one TD, one pick) didn't look sharp except for a couple nice throws on roll-outs, but he always delivers the fantasy goods (44 rush yards, two rush TDs.)
• Miles Sanders (9-56-0, 4-2-10-0) wasn't used much, but showed his speed and explosiveness when he was. He also got one goal-line carry, but was stopped at the one.
• Zach Ertz (6-4-29-1) salvaged his day with a TD, but no Eagles receiver did much. If he were a real man, he would have suited up Sunday for the Cardinals, gotten dealt to the Titans and played in the Monday night game too.
• I loved the Eagles going for two down eight in a game with a seven-point spread. If they miss, the cover is probably good, and if they make it, the Bucs have more urgency to kick a field goal to make it nine and not risk losing the game. But stupid Brady and his QB sneaks.