Data scientist, physicist, and fantasy football champion

Week 10 Flex Predictions

I keep trying to automate Flex rankings so I can prove I’m better than everyone else. So far it hasn’t worked out great. Week 8 I would have slightly beaten FantasyPros ECR, but week 9 I definitely would not have.

I really didn’t like the nonsense my model put out last week, so it’s not even worth it to post. I did slightly worse than FantasyPros ECR. It was really nonsense. But let’s try again and see how my flex rankings look for the week. We’ll start with the top-36:

.graph mix model A-1.png

Now THAT’s a top-36. The only guys up here that worry me are guys on my team who did poorly last week (Looking at you, Aaron Jones and Doug Martin). I don’t think there’s a guy on this list that I wouldn’t put into my starting lineup. And even my TEs look good: Gronk, Engram, and Reed (if healthy) are all good plays with the potential for big games.

Off to a good start this week, model. Let’s see the next 36:

.graph mix model B-1.png

[Again, if you’re in a 12-team league, you should hope you’ve seen all of your starters in this list or in the first list. Most leagues have 3 WR / 2 RB / 1 TE or 2/2/2, for a total of 6 WRT positions. 6 times 12 teams is 72 players, so ideally all of your starters will be in the top-72, right? Right.]

Again, a lot of great picks here. Delanie Walker gets enough yards to be buring somewhere in this list. He’s a little higher than I’d think, but as the 4th TE off the board I really like him. I like Kenyan Drake and Damien Williams around here as well. Last week the model didn’t know what to do with them, but I think this looks about right to me.

The guys I don’t like are guys with a fairly recent change of situation. Will Fuller is probably higher than he should be given that it’s not Tom Savage throwing to him, but the model tries to take that into account which is the only reason he’s down here and not in the first list. Davante Adams is tough to put as the 55th guy overall given Brett Hundley’s performance. Wayne Gallman is also a tough sell since they seem to be going with Darkwa for now. Same thing with Austin Ekeler (backup to Melvin Gordon). Those two are a case of my model seeing guys who have scored a few points in the past going up against exceptionally soft defenses. Then again, they’re on here as the 30th and 36th ranked RBs off the board, which is entirely possible for them. I think my problem is less with their ranks and more with the fact that I’ve labeled my y-axis “predicted fantasy points”. If I just changed that to “player confidence” or something I’d be a little less nervous about them.

Oh, and I really don’t like Giovanni Bernanrd up here. Get out of here, Gio! Sorry everyone, I don’t know how he snuck in here. I need a way to model players’ floors and ceilings to show how I really feel about a few of these guys.

I really like of the rest. Let’s see the guys who just missed the top-72:

.graph mix model C-1.png

Cool. If someone asked me to create a list of medium-to-high-ceiling/absolute-sub-basement-floor guys it would look something like this. Your Kennys (Britt, Golladay, and Stills) are all fairly high ceiling guys. Decker has been amazing in previous years. Hurns has had a TD or two.

Goodwin and Funchess are the inverse to guys Davante Adams and Will Fuller, that is, guys in a new situation (losing the WR1 ahead of them) and the model doesn’t fully account for the change. Still, I’ve proven to myself (in an article I never published) that losing the WR1 ahead of you doesn’t always instantly make you the WR1; it tends to make you a (WR2 + a few points).

I like about half of the TE options, but I never really know what to think of TEs. With the exception of Gronk, Ertz, Kelce, Reed, and Engram, they tend to all get 10-20 yards with a few random TDs thrown in. For all I know Adam Shaheen (TE-CIN) might really be a good pick this week. Against… let’s see… GB who have given up the fewest yards to TEs all season. But Zach Miller’s out, so… Ugh, who even knows with these guys.

One last graph:

.graph mix model D-1.png

A handful of guys with upside that I’d be willing to play as a flex if needed (Charles Clay, Josh Doctson, Breida, Kroft, Celek), but mostly guys that are best avoided.

So that’s week 10 Flex. I really like the looks of the model this week. Most guys make sense and while I have (capital-O) Opinions on some of the players I think my model will do well against FantasyPros ECR this week.

I still don’t trust it, though.

Week 10 DEF Results

Week 10 QB Predictions