Data scientist, physicist, and fantasy football champion

Week 4 DEF results

Again, I followed my own advice (with Cincinatti), And again, they were solid but unimpressive. It seems I can’t predict TDs, I can only predict mid-to-high performances. I guess that’s okay? I don’t know. My DST choice has yet to win or lose my week, so that’s better than last year, but my models are still unimpressive.

Let’s see how each model did last week:

.accuracy graph-1.png
Modelestimatestd.error
ModelA0NA
ModelC-6.81256.367158167
ModelF-5.56256.367158167
ModelG1.43756.367158167
ModelH17.48.934987409
ModelYahoo-11.06256.367158167

The new model (H) was terrible, and Yahoo was the winner. In fact, they’ve been looking pretty good this season. The table above shows the average deviation from model A, i.e., with negative scores meaning better accuracy. It looks like Yahoo is the pretty clear winner so far this year with Models C and F pulling in after it.

I may actually use Yahoo’s experts’ predictions this week. Damn. That’s hard to say out loud. Maybe I’ll keep trying Model C. Using other people’s predictions feels dirty.

Week 4 QB Results

Week 4 QB Predictions