Data scientist, physicist, and fantasy football champion

QB points by rank

QBs were weird in week 1. Brady, Brees, and Rodgers were all outside the top 12. There were 6 QBs with negative scores (by our rules which penalize QBs 1 point per sack), most notably Andy Dalton who just… I mean… damn dude. The top 3 QBs were Alex Smith, Matt Stafford, and Sam Bradford. Outside those top 3 it seemed like everyone else clustered around medium to low scores.

So week 1 was just a statistical outlier, right?

Or was it?

Let’s see how week 1 fantasy QBs stack up against the regular seasons for the last two years:

.weekly pts by rank-1.png

OK, yeah, week 1 was bad, but not necessarily an outlier. Yeah, the last few QBs were pretty awful, but the middle wasn’t that much lower than the median QB performance. In fact, let’s see that median:

.median pts-1.png

That is fascinating. The top QB in a week has a median around 31 points and the 32nd ranked has a median of about 0. You can pretty much write:

[Fantasy Points] = 32 - [Rank]

and you'll get pretty close. To be fair, there’s maybe a little more structure than that, with a sharper dropoff for the first 5 QBs, a shallower drop over the next 20, then it falls slightly faster again after about rank 25. Still, really neat trend. Of course, this is just the weekly totals. Check out season long ranks here.

I don’t yet know how this will affect my decisions week-to-week. If I look at my models and they put Matt Stafford at 12th and Carson Palmer at 15th, will I choose Stafford for a little less than 3 extra points? Maybe in a vacuum, but what if I don’t want to drop Palmer because Stafford has better matchups the next few weeks? Is it worth it to potentially lose a player I want to hang on to over less than 3 points? If I’m pretty sure that neither of them will be in the top 3 this week, then is there enough difference between them to justify making a switch?

This isn’t just academic. I have Stafford. My models put Smith, Flacco and (sometimes) Palmer in the top 10 this week, but Stafford closer to 17. That’s about 7 points, which is no joke in my league where the weekly expectation is around 100. I want to stream one of them this week, but I don’t have a ton of bench room and I hate wasting it on QBs. But I don’t want to straight-up drop Stafford after a monster week 1 performance. But if I can always stream a comparable QB, then why am I holding him? But he’s had a few monster seasons, so what if he’s actually worth it to hang on to? But Beyonce told me to not ever get to thinking he’s irreplaceable.

I’ll probably clear up some bench space, stream one of these guys this week, then just drop them next week for whatever hot new RB prospect is out there.

Does in-division affect fantasy scoring?

QB reliability (a.k.a., A note to Marcus Mariota and Matt Stafford owners)