The NFL kickstarted with an absolutely thrilling week 1 slate, with almost every game decided by one score and two overtime games! Unfortunately with that came a lot of upsets, and a LOT of unexpected results, which led to poor model performance. The overtime games really changed the outcome of this week, as if both kickers had made their respective game winning field goals, the model would have been positive…

As mentioned multiple times, the model does not really start beating sportsbooks till around Week 4, so this is nothing unexpected; furthermore, Week 1 has been the poorest performing week in all 3 years I have tested the model.

Every week I will tally up both the YTD results since the beginning of the season as well as the current week of stats. Since the NFL just began, the cumulative and Week 1 results will be the same.

Week 1/YTD Results:

Total Units Staked: 7.9 units
Total Units Won/Lost: -3.47 units

ROI (version 1):

I will have two ROIs shown for every post. The first will be in terms of units staked, version 1:

(total units won/total units staked) * 100

The other version will be in terms of the total bankroll, version 2. For this, I will assume each unit corresponds to 1% of your bankroll; if you used 2%, you would multiply the version 2 ROI by 4x.

(final bankroll-initial bankroll)/initial bankroll *100

Version 1 ROI: -43.9%

Version 2 ROI: -3.47%

Overall, the poor performing week did minimal damage to the total bankroll since the bets were scaled down so low. In hindsight it would have been smarter to scale down the bets from week 1 even more given the knowledge that it is unlikely to perform well till week 4, but this is a lesson to take into next year!

An important thing to remember is now since my total bankroll is down 3.47%, my unit size for each bet will change (and should for anyone following along). So if you started with 100$, a 1 unit bet would go from 1$ to 97 cents if you chose your unit size to be 1%.

Again, remember that this is likely the worst performing week so this is nothing to be concerned by. Week 2 performance should drastically improve as there is now current season data to feed into the model, so stay posted for Week 2 predictions!