Week 1 2020 picks
Nature never repeats herself in quite the same way. “The Tyrannosaur doesn’t [abide] by patterns or park schedules. The essence of chaos…is dealing with predictability in a complex system.” And so, in Anno Domini 2020 the cloudy, VUCA NFL environment returns with a vengeance the size of T-Rex…
Will games be postponed or cancelled? Will fans return to the stands this season? Will Tom Brady and Drew Brees live longer than Noah in the Old Testament? These are questions we want answered! Well, that and which teams will cover the spread each week.
Seriously though, if you’ve been a loyal reader of this blog, you’ll know that in times of uncertainty the best course of action is to stick to the process. So, in a lot of ways, the method to get the algorithm in shape for this 2020 season was just like every other: evaluate team performance to close out the prior season, regress team rankings toward the mean, adjust for under/over performance using a modified Pythagorean formula, tweak results for offseason player/coach movement, account for market movement and voilà! That was easy!
Since I missed the opportunity to share the pre-season power ratings last season, as the Algos For The Everyman blog debuted mid-season, I thought I’d outline where I rate the NFL teams going into this unprecedented 2020 season.
Group 1 - Super Bowl Contenders
BAL - 25.7
KC - 24.2
DAL - 23.5
SF - 22.3
NO - 22.0
MIN - 21.8
Group 2 - Playoff Caliber
TEN - 21.8
IND - 21.7
SEA - 21.0
TB - 20.9
BUF - 20.4
GB - 20.2
Group 3 - League Average
LAR - 20.1
ATL - 20.0
LAC - 19.8
PHI - 19.5
NE - 19.3
Group 4 - Not This Year
ARI - 19.2
DET - 19.1
PIT - 18.9
HOU - 18.8
DEN - 18.4
CHI - 18.1
LV - 18.0
Group 5 - Plug Your Nose Contrarian Plays?
CAR - 17.3
CLE - 17.0
NYJ - 17.0
CIN - 16.2
MIA - 15.9
NYG - 15.8
WSH - 15.0
JAX - 13.5
————
The Tableau visualization on the cover page of The Picks shows the forecast variance across the other models I track as well as where the contest line lands in relation to my raw algorithm prediction and the “average” of all the prediction models, including the Vegas line. If you change the filter drop-down menus to prior seasons, you’ll find variance reduces as the season progresses. Thus, it should come as no surprise this Week 1 forecast range is very large. I’ve also observed process-based selections seem to outperform the consensus in the first part of the season. Having reflected on this observation, I’ve concluded the best explanation might be that using an algorithm to ground your weekly selections proves an effective antidote to the recency and availability biases that are particularly virulent in small sample size assessments, like Weeks 1-6 of the NFL.
For those of you reading for the first time, I’ll provide the raw algorithm suggestions with their relative strength as denoted with a “★”. I’ll bold the top teams the algo has identified as a potential play after analyzing the sharp/contrarian sides.
For reference, here’s the 2019 ATS performance broken down by algo strength:
★★★ = 64%
★★ = 51%
★ = 47%
With that behind us, without further ado here are the picks for Week 1 of the 2020 NFL season.
Raw algorithm picks:
CAR +3 ★★★
NYJ +6.5 ★★★
WSH +5.5 ★★
ATL +2.5 ★
NYG +6 ★
Sharp/Contrarian Sides:
HOU (@KC) - Moderate - KC, fresh off the SB victory and with the better offseason, it’s no wonder they’re 9 point favorites in the contest. The line opened 9.5 and has shifted ever so slightly towards HOU. Raw algo shows HOU +7.2 in this spot and I’m liking HOU here.
MIN (GB) - Weak - I think I’ve handicapped this matchup well. Raw algo shows MIN -3.1, line opened at MIN -3, and has shifted towards GB +2.5. We know 3 is a key number and GB is a public team, so not surprised for the move off the key number, given the public is backing GB at a 2:1 clip (on Wednesday).
IND (@JAX) - Weak - Line opened IND -6.5 but is offered in the contest at -8. A ticket split of 60/40 indicates the -8 might be closer to the appropriate number even though my handicap lies closer to the opener. I'm noting the sharp side on IND but I’m leaning with JAX here.
NYJ (@BUF) - Weak - Line here opened at BUF -6 and has moved to -6.5 off 71% public action. The algo likes this play but the honest 0.5 point move following the public tickets gives me pause. It’s a possible play but I’ll likely wait for any weekend late-line moves before getting in.
WSH (PHI) - Strong - Classic sharp move here. Public all over PHI—line opened PHI -6—but has been moved to -5.5. Given my algo is also on WSH, I’ll be making this play.
ATL (SEA) - Moderate - I get a home ‘dog, with a darkhorse playoff team, coming off a scorching 6-2 record to close out the 2019 regular season? I’ll take it. Line opened ATL +1 but has shifted in response to heavy ~3:1 public action on SEA to ATL +2.5.
SF (ARI) - Weak - My algo likes ARI here but SF seems like the better choice. Public on ARI 2:1, no discernible line move, hard to take 6.5 in this spot. Pass.
NYG (PIT) - Moderate - The only reason I’m even considering this is the “turn your gut” rule. That is, the game(s) that you feel are definitely the wrong side are the ones that end up being right. Line opened PIT -3.5, which, in a vacuum, seems fair. Until noting the algo has this game at basically a pick ‘em and the public wants nothing to do with NYG since the contest is offering NYG +6. Plugging my nose and considering the number here.
CAR (LV) - Moderate - I must be missing something with Carolina this season. Teddy Bridgewater is 16-6 in his last 22 games and they have Christian McCaffrey in the backfield. Nonetheless, the algo has this game as a pick ‘em, whereas the public (2:1 tickets) has moved the line from LV -1.5 to -3 in the contest. The full FG line seems like fools’ gold in this situation but I’m feeling CAR as a home ‘dog.
DEN (TEN) - Moderate - The Broncos opened as 2 point favorites, but that’s been flipped on 70% public Titans tickets. I see myself in the public here so I’m leaning towards the Broncos. The raw algo shows fair value on DEN +2 but the contest has them getting another 0.5 point.
Unofficial plays: