Fantasy: Broke Models Need You

Sorry to disappoint- this isn't that kind of article. Get your minds out of the gutter and back on fantasy football.

Fantasy football player values are unknown. In every discipline, from biology to economics, we build models to predict unknown values. Then we hope they work.

You just have to Google “fantasy football” and you'll get a long list of sites with models that predict future fantasy football points. They've been created by clever—and not so clever people to sell advertising space that you'll mostly ignore.

They're all wrong. But pick one anyway. CBS Sportsline will do well enough in a pinch. The model that you use matters much less than how you use your model.

All of the fantasy projections are far too dependent on statistics. Some are too dependent on last year’s statistics, others are too dependent on this year’s statistics—but all are based on running stats through some formula.

Statistics lie. Consider this stat line from last week's Jacksonville- Denver game:

REC YDS TD FAN
Nate Jackson TE (DEN) 0 0 0 0

Goose eggs. That isn't the whole story—if you watched the game, you know that Nate Jackson dropped a pass in the end zone. One catch for three yards and a touchdown is a pretty good day for a tight end.

If Jackson is an end zone target for the Denver Broncos, most models aren't going to price that correctly, they’ll just see the zeroes.

Models also handle injuries very poorly. They tend to under-value the negative impact on the players around the injury. I think the models anchor themselves to their previous projection.

So if both the projections and human beings tend to make this mistake, you may find opportunity by trading away players on teams that have a newly injured player.

All the Dallas Cowboys players went down sharply in value on Monday when we first heard about the pinkie, rebounded Tuesday with the Roy Williams signing, and rebounded further with Romo maybe playing hurt.

How much? I don't know, but I can tell you which direction they moved in. Sometimes that's all you need.

If you're more aware of the current events changing the marketplace than your opponent, you should be able to create value for yourself both in trades and on the waiver wire.

Know when your model is broken, and you'll make better decisions.

Edited By Matt Gilmartin