Walking Blindfolded or Seeing Clearly?
Last week, I questioned the Dolphins pick but went with it. My model ranked Miami 5th of 16 games, and just a shade above our filter limit.
(Reminder: filtered means, for a pick to be "worthy" it must be far enough away from 0.5 in the model's prediction. Consider 0.50 a coin flip at the given spread or total and the further from 0.5 you get, the more the model has confidence in the given choice. ATS limits must fall outside 0.45 / 0.55 while Totals are set at 0.43 / 0.57).
Regardless, Miami was our only ATS loser thus far (14-1 ATS Overall, 10-1 Filtered, 3-0 on Top Choice), while Totals are hanging on by a thread at 8-7-1 on model filtered picks (1-2 on Top Choice).
This week, the model has a Thursday night pick, GB -4, as the second rated of the week (Phi @ GB -4) with the huge favorites (anyone playing Miami) as the top rated (LAC @ Mia +15.5). My gut says that GB should in no way be giving more than 3, so what am I missing? If I follow this pick, am I walking blindly by following the model, and ready to find the cliff of regression after a great 14-1 start? OR, am I finally seeing clearly by using a quantitative model (exactly what I set out to do with this model).
Going back to last week, we took Pit last week after Roethlisberger was injured and the known starter was Rudolph on the premise that Big Ben played below his normal standards thus Rudolph coming in may not be that much of a change and could even be a jolt (he was not, Rudolph was terrible and I felt t lucky having Pit cover given a +3 turnover ratio at fortunate times).
Back to this week, Philly had a lot of injuries lately and easily could be the reason they lost straight up and ATS in their past two, however, with a team stat based model, player injuries (or their recent bout of WR drops) are not directly accounted for, only indirectly in you consider the market accounting for them via the spread and total.
Regardless, without a major injury or other situation, I am trusting the model and moving forward.
This week, only the top two are filtered, but like always, I'll give my top five if I was forced to take exactly five, like in the Super Contest (using Westgate's Super Contest lines for Week 3).
1. LAC -15.5 (LAC @ Mia +15.5) - Top ATS Pick, Filtered
2. GB -4 (Phi @ GB -4) - Filtered, *Thursday night
3. Was +3 (Was @ NYG - 3)
4. Chi -1.5 (Min @ Chi -1.5)
5. Det +6.5 (KC @ Det +6.5)
It is hard for me to qualitatively take Detroit, as I watch them every week, and their best CB (Darius Slay) has a hamstring issue / limited practice participant. However, Detroit can score and has potential for a backdoor just like Baltimore did to KC last week.
In model we trust (said with a calming exhale that only masks the stress of betting against KC).
If I had to choose betting ATS or Totals or only able to win at one of those two, I'm taking ATS, so overall I'm still thrilled at the how the season has gone: having ATS do so well (albeit a small sample size) and Totals hovering around 50%.
As we eluded to above, it is a small sample size and without a major reason to disagree with the picks. Regardless, I will be looking to tweak the Totals model now that we have some 2019 data, but no changes yet. Thankfully, our "Poor" performance is still a game above 0.500, at 8-7-1.
The ATS model only chose two games this week, however the Totals model has NINE!
All nine of there are "filtered" (my way of saying it is worthy of a bet):
1. LAC @ Mia - Under 44
2. Ten @ Atl - Under 45.5
3. Was @ NYG - Under 49.5
4. Car @ Hou - Over 47
5. Cin @ Pit - Under 43.5
6. Oak @ Ind - Over 45
7. Phi @ GB - Over 46.5 *Thursday night
8. Dal @ NO - Under 47
9. Sea @ Ari - Under 48
The model loves has tended to lean away teams and unders, so when home teams and overs come around, it is worth noting.
Have a great week 4.