Week 1 Pick’Em and Projected Game Scores
Double the Content, double the fun
It’s been a busy off-season. I will be keeping the Medium posts short due to the manual nature of the posting process.
I have a similar post on our site, check it out if you have time https://gridironai.com/football/blog/article/week_1_2019_pickem/20
Backtested Performance Results
Until I have time (and fully backtested results) to write a performance summary, below is the current backtested results, by season for the Pick’Em predictions. The backtested results (below) are done by simulating the week to week predictions but this hasn’t been tested in the wild.
Transparency — The far-right column seems too high. If we average 70%–80% this year, I’d be highly surprised. I attempted to find issues with the data that would cause it to predict this week (known as data leakage) but I could not. We’ll see how it shakes out in the first few weeks and go from there.
Current Limitations
Below is a list of the current known limitations in our model:
- The model does not know about major roster changes. For example- It doesn’t know Luck isn’t starting in Indy anymore. I am working to get this information added
- It hasn’t been tested in the wild, meaning, I am going to have a good idea if I messed something up after the Sunday games
PICK’EM
The current model is predicting the following information:
- Home Team Score
- Away Team Score
- Vegas Line — In general, this should track pretty closely to what Vegas is set for a game.
- Prob HT Winning — Probability we think the HT is going to win
Notes
- Bears lost, that sucks. We did have it at 54% which means it wasn’t egregiously wrong
- Panthers are low because it doesn’t know Newton is back, so that is less of a lock than it seems
Week 1 Projections For Fantasy
Check out the post on our site if you want AI-generated projections for Fantasy Football!
https://gridironai.com/football/blog/article/2019_week_1_projections/19
