In NFL Week 8, the First TD Model had its first losing week of the season, but that's expected to happen as betting first touchdowns is quite volatile, and the goal is long-term profitability on these.
If you were to bet every single +EV edge the model picked up on its final update before each game after inactives came out, using quarter-Kelly bet staking, you'd be up the following amounts:
- Week 6: +25.33 units
- Week 7: +0.19 units
- Week 8: -3.96 unit
- Overall: +21.56 units
Each week, the dashbboard will be updated daily, with additional updates after inactives are announced an hour and a half prior to kickoff of each window of games.
I'd strongly advise to wait for inactives to come out before firing any bets. The midweek version of the model can be a bit off as practice reports come in and change things, and as players get taken off or put on IR it takes time to update the data. Treat the post-inactives update as the final and best version of the model for each game.
Here's the Week 9 NFL First Touchdown Model, starting with Jaguars vs Ravens on Thursday Night Football!
Also, I've added a few other features to the model, namely:
- Quickslip links: Under the "Bet Now!" column, right click the logo of the book or prediction market and select "open link in new tab" to be taken directly to that book or market with the bet slip pre-selected for the player you'd like to bet.
- 95% Confidence Interval: If you're worried about how accurate the model is, I've put a toggle on the board to just look at the low end of the 95% confidence interval for each player's first TD chances, to help filter out any that barely clear the +EV threshold which fall within the range of noise. That said, the model default is well-calibrated and has proven succussful so far.
- Kelly Criterion: I've added betting sizes based of the model projection (or the lower 95% confidence interval projection) based off the Kelly Criterion, defaulting to one-quarter Kelly sizing to help reduce volatility and account for model error.
