Yeah I agree.. was just pointing out the only actual statistical evidence you'll find to make an argument for who is more indispensable favors Manning. Take away the stats stuff and I'd say an in-prime Manning is more indispensable because I feel like he makes an OC less necessary than an in-prime Brady. Oh, and I absolutely agree on the "Suck for Luck" thing being very suspicious with Indy. They only won games after you could argue it almost didn't matter (they started 0-13 and could easily come out with the #1 pick by losing one more game).
IMO, their defense was much better in 2008. With their defensive losses this offseason, if Brady goes down, they are 9 win team
I'll give you props for asking a good question, but even if one took into account opponents' records, you'd have problems. You would somehow still have to weight the opponents' records by their opponents' records, and so on and so on. You'd also get a confound in that two of the opponents for each division team would be the team you're trying to analyze (NE vs. Colts). The proper way of analyzing this isn't to look at opponents' records. The proper way is a game-by-game analysis where you start all teams off with the same points, and adjust those points based on the result of each game. The math to use would be ELO rankings. You can either do win/loss only or take into account win differential (so, points scored) if you know what the distribution of win differentials on average for NFL teams are. ELO is the same math used for example to adjust the rating of chess players after each match they play. Anyway, I am obviously not going to go that far for a post on a message board So how close to a final ELO ranking would just looking at the records be? I'd bet they're not too far apart given how the NFL creates schedules. Also note that the way the NFL determines strength of schedule in the next year is exactly by doing what I did, which is just looking at win/loss records. So in some sense what I did is more "official" even though it's not as good as a game-by-game analysis. Having said all that.. we can go ahead and look at the records of common non-division opponents (so the records of teams in the common AFC division and the common NFC division) in the 4 cases: win % of all common non-division opponents of the AFC East in 2007: 72/128 = 56% win % of all common non-division opponents of the AFC East in 2008: 45/128 = 35% win % of all common non-division opponents of the AFC South in 2010: 63/128 = 49% win % of all common non-division opponents of the AFC South in 2011: 70/128 = 55% So looking at common non-division opponents paints the same picture! NE without Brady had an easier schedule, both in absolute terms and relative to the year before, than the Colts did without Manning. So statistically speaking, Manning is more indispensable than Brady for the only year you can use stats.
In 2014 the entire back seven w the exception of Grimes missed multiple games due to injury or suspension. Coyle was bringing guys off the street to start at CB. Jim my Wilson started 4 games at safety. Finnegan, Thomas, Wilson, Davis, Taylor, all injured. Ellerbe played one half of one game. Misi was hurt all year. Wheeler and Jenkins both got hurt. Delmas tore his ACL.That's just the defense. Moreno played one full game. Albert missed half the year. Pouncey out four games.
Ahhh you used the golden ELO ranking!!! I'm very interested in the ELO ranking being applied to the NFL. I've always wondered why its never made the jump. The methodology you're using is sound. I just don't agree with the interpretation of the numbers. It feels like there is something missing to the equation. Did you mean NE didn't have an easier schedule? I'm lost.
Some people have already applied it, but it hasn't caught on yet. For example: http://fivethirtyeight.com/datalab/...gs-and-playoff-odds-conference-championships/ Your more general point is still correct: it can take an excruciatingly long period of time before a well-known method of analysis is finally applied to a field. The reason for that is there is so little communication among fields that people in one field don't know there's a long-since-discovered better solution to their problem in another field. The lag in implementation is often measured in decades. I'm open to suggestions.. not sure how else to interpret the numbers though. What's missing of course is all the information I didn't include (like non-common opponents' records). Question is how likely is that info to change the overall result? Probably not much. The win % of all common opponents of AFC East teams in 2008 - the year without Brady - was lower than the win % of all common opponents of AFC East teams in 2007 - the year with Brady (35% in 2008 to 56% in 2007). Remember, this is win % OF the common opponents, not AGAINST the common opponents (you asked for opponents' records). This means NE without Brady had weaker common opponents in 2008, because their win % is less. For the Colts, the result was the opposite: their common opponents without Manning had a higher win %. So NE had an easier schedule without Brady, while the Colts had a harder schedule without Manning.
