For a wiki reference for any terminology that is familiar, refer to this article.

To preface this week’s results, I wanted to lay out some more changes, as I will continue to try to make this process more smooth over the next few weeks. Firstly, upon feedback, I will post the Thursday night and Monday night games on Thursday/Monday morning at 6 AM MST; although the odds may be slightly different, this should be negligible unless there is a large injury. If there is one, I will post an update within that article 1 hour prior to the game. I will also make a post regardless of whether there is a model bet, and if there is no bet the article will just say “no bet”, also to make it easier on the user. I will post the results Wednesday night at 7 PM MST from now on. I will also be adding the parlays to the excel sheet to make everything easier. Finally, now that there is enough of a sample size, I will also be posting the results in graphical format starting next week in addition to the current format, to track both versions of ROI over the course of the season (both in terms of YTD and YTD Starting at Week 4).

Now that a week is in the books, the model has shown how much/many bets should be expected, and the user can manually scale down either the unit size (not recommended) or weight down the bets equally (this is the recommended option) based on how much they choose to risk in a given week. This can easily be done In the weekly excel sheet under the “user inputs” table. I will also be scaling up the parlays as they already seemed to have the biggest positive effects from the week (~+6U!).

Although I was hoping for very positive week, it was basically “meh”, with this being the 3rd big bet in which a kick under 20 yards cost the model big (looking at you Greg Joseph!). I plan to post some more insights into the model overall hit percentage on my test sets/progression over an nfl season of hit percentage next week to give users a better expectation/make judgements for themselves.

Despite the “meh week”, there are some interesting points here. Firstly, Weeks 1-3 will continue to be more and more negligible over time since the bet sizes (except parlays) were and will be at full scale now (meaning I am outputting directly what the model is doing without scaling it down). Overall, it was still slightly profitable, so there is no complaints for the week!

Week 4 Results

total units bet: 44.11

Total units won/lost: +.49

ROIv1=0.88%

ROIv2=.49%

YTD Results

Total units bet:76.14

Total units won/lost: -5.01

ROIv1 =-6.57%

ROIv2: -5.05%

Remember there will also be a YTD Results Since Week 4 starting next week. Stay tuned for next week’s picks!