Week 9 2020 picks
Though we missed our Week 8 post to attend to some joyous personal matters, it was a banner week for both the algorithm and the contest entries! Entry A went 5-0 and the raw algorithm went 4-1! Last season, Entry A was chugging along very nicely through 8 weeks only to throw in a 0-5 right smack dab in the middle of the season. That prompted me to write a reminder to myself to "follow the process” through thick and thin, when recent results and confirmation bias are making it seem like I’ve got it all figured out to when the outlook is so cloudy I can’t seem to make sense of anything!
It’s also tempting to stray from the process to make a context-specific handicap—that is, to use a popular external narrative to influence your prediction. This is a cousin of availability bias. That’s the bias we succumb to when we have information readily available to us but choose not to validate it’s cue. As the availability of information becomes ubiquitous, at the touch of our thumbs even, we need to implore our System 2 to put forth the effort to widen our mind’s aperture, to seek countervailing opinions. Context is sexy because it’s often the thing that those on either side of an outcome end up pointing to, disproportionately I might add, as causation.
For NFL games, context comes in two forms: internal and external. Internal context has a direct bearing on the outcome of a game. This context is best articulated in the NFL Next-Gen Stats Amazon AWS commercials and challenges those in the NFL’s Big Data Bowl to find predictive power from actual game context (player coordinate positioning, personnel, game situation, etc.). In the most sophisticated models, like Football Outsider’s DVOA metric, this internal context influences the weight of model attributes.
One of the shortcomings of the model I’ve developed is that it largely discounts internal context when applying a post-game power rating adjustment. For games with a lot of garbage time scoring, my algorithm could misjudge the game flow. However, if we recall the Jurassic Park clip I shared in the Week 1 opener, just like water flowing down her hand doesn’t happen in quite the same way each time, no two NFL game flows are quite the same. So even accounting for every detail of a game still has limited predictive power.
Now what the NFL pregame shows feed their viewers is almost exclusively external context. Stories like Russell Wilson’s mental conditioning, the sad state of affairs in the NFC East Division, whether Tom Brady with Antonio Brown can avenge a Week 1 loss to the Saints. These are all examples of external context. Does Russell Wilson’s mental conditioning help him every game? Absolutely! Will it be the differentiator in Week 9? Likely not. Are the Cowboys even an NFL-caliber team anymore? Eh, not really. Does that mean they won’t cover 14 at the Steelers this week? Well no, not exactly. I have started to notice a fade-trend occurring with the teams profiled on Fox NFL Sunday. The team(s) profiled (read: provided viewers a heavy dose of external context) end up not covering. I’ve casually observed this, but will attempt to validate this in the offseason, time willing.
This isn’t to say that all off-field context is irrelevant. Injuries should definitely be taken into account when handicapping. However, without tempering expectations for how much the available external context should affect the game-at-hand, you risk falling victim to availability bias.
————
Here’s a quick recap of the entry performances for the last 2 weeks:
Entry A (Alex): 7-3
Entry B (Sean): 5-5
Entry C (Bryan): 5-5
Entry D (Alex & Erik): 5-5
The top-25-ish players in the contest averaged 5.5 over that span, so save for our first entry, the rest of the group pretty much treaded water.
Now for the Week 9 games…
Raw algorithm picks:
DET +5.5 ★★⚐
MIA +5 ★★
NO +4.5 ★★
CHI +5.5 ★★
DEN +3.5 ★⇩
Not that any week is “easy”, but there are some that seem set up for a high scoring week and others where just getting 2/5 keeps you in the game. This week feels like one of the latter types. Value seems to me like a mirage…like the top 4 algo plays this week. My gut likes the other side in all of those, which I’ll get into as we review the sharp/contrarian sides...
Sharp/Contrarian Sides:
SF (GB) - Weak - Boy are the Packers tempting to take this week. The ‘Niners are missing a ton of players while the Packers are ripe for a get-right game. The algo favors GB by about 3.5 so laying 6.5 in the contest…on the road…on short rest, doesn’t seem like a profitable betting strategy. On the other hand, playing SF (whose backup QB I can’t recall) against a hungry Rodgers, also doesn’t seem appetizing. The question is, and it’s a valid one, has the public overreacted to all of this external context? The public ticket count is 2:1 on the Packers, while the money appears to be rolling in at a 5:1 clip. The ‘Niners are obviously the contrarian side here but I would’ve needed to see 10-15% more public ticket lopsidedness before making it a suggested play.
CHI (@TEN) - Weak - The contest is offering this game at the Titans -5.5 at home. After losing last week, they’re primed for a bounce-back game against a formidable Bears team. Ticket count is dead-even as of this writing. My line makes this more like a FG in favor of the Titans so some value appears on the Bears. However, my gut (and this will be a theme this week) thinks TEN is good to cover. The market line opened and has remained at 6.5, which counterintuitively supports my thinking here. The Bears are the contrarian pick but I think the Titans are the right side.
DET (@MIN) - Moderate - Similar story as above for this game. The contest has this line at -5.5 favoring the Vikings, the market line opened and seems steady at Vikings -4.5 with a split even ticket count. Given the market line is slightly favoring the Lions, there might be some value to be had on this one. My algo thinks this line should be closer to 3.25. All that said, I do think the Vikings win this game outright, and we know that the winning team covers 80% of the time, so when placing a wager on the Lions, you’re really saying is that you’ll hope for a thread-the-needle-type game. Lions are the play here. Good luck!
BUF (SEA) - Moderate - Again, here’s a game where I see the interest in the Seahawks off of a solid win against kind of healthy ‘Niners last week. I think it’s slight public overreaction. The question is, is there enough of an overreaction to make the Bills a play. The contest has this line at Buffalo +3, while I think +1.5 is fair. Backing my algorithm is the market opening number of Seahawks -1.5, telling me my handicap is calibrated well. The fact that the public is all over the ‘Hawks at a 2:1 clip, laying points on the road, tells me the Bills offer slight value here.
IND (BAL) - Strong - This is the type of sharp line move I love to see! This line opened favoring the Ravens by 3 but has been adjusted down to 1, even though 60% of the public tickets are on the Ravens! The contest has this line at Ravens -2.5, while my handicap is essentially 0 for this game. Strong play on IND here.
CAR (@KC) - Moderate - To no one’s surprise, the Chiefs are the highest power-rated team in many models and with a bye on the horizon, this is an interesting spot to keep momentum into the off week. However, I like Andy Reid coming out of a bye week rather than going into one. The Panthers are no pushover and getting 10.5 in the contest versus my algo suggested line of 10 shows slight value with them though not enough to push them into my top 5.
JAX (HOU) - Strong - Here’s another sharp move, though not quite the same magnitude. Both teams coming off a Week 8 bye and both in the bottom third of my power ratings. The Texans winning this one outright seems like a foregone conclusion, but laying 7 in the contest, when the fair number is closer to 3-4, even when adjusting for the Jags’ rookie QB’s first start, seems like a play. The market actually has moved towards the Jags as well having opened at +7 and moved down to +6.5 even though only 44% of tickets are on them. External context warning here, but I like the rookie QB with the added prep time coming off the bye week to cover in this one. Jags are the play.
DEN (@ATL) - Weak - Algo is all over the Broncos in this one (fair value at a pick ‘em), but after squeaking out a win against the Chargers and now having to travel to Atlanta, who are surging in their own right, it seems like the gut play, the Falcons, is the right play. The line opened at the Falcons -4 and largely has steadied there, save for some books, and the Falcons are getting 55-60% of the public tickets, which feels about right. Some would say, “this is why you have an algo, to plug your nose and play it” but I’m going to hold off.
DAL (@PIT) - Moderate - You don’t need to look at the market line to know where the value is in this matchup. Clearly America’s Team no longer inspires confidence against the number so this line’s been inflated to two touchdowns (PIT -14). On top of that, the public is backing the Steelers at a 70% clip. Which might be the most one-sided public opinion in a few weeks! Forced to play this game, I’d take the ‘Boys, but since we must only play five games, I’m going to hunt for value elsewhere.
LV (@LAC) - Weak - Following the theme this week, game context, the question that pops out here is, how much do we expect the Chargers to play up to their potential following, yet another, lead evaporation? The algo likes LV here by about a point and that’s where the market is headed. The line opened Chargers -3 but has been bet down to LV +1. The contest has this game at a pick ‘em and I think that’s where we’ll get by kickoff. Question is, are we underweighting the Chargers actual potential here based on the raw results? I think so and for that reason I’ll follow the money at take the Chargers.
MIA (@ARI) - Weak - The Cardinals are getting more of the sharp money it appears, though tickets are pretty split between these teams. The market line opened -4 and has moved up to -4.5. The contest has this line at 5, while my algo suggests fair value is closer to a FG. Once again, my gut agrees with the market move but we can’t discount the algo. While I do like the Cardinals to win outright off a bye, I do think 5 has some value here.
NO (TB) - Weak - Classic rematch of their Week 1 showdown, this time Brady has full control of his new team and appears ready to even the season, head-to-head tally. The market opened at the Bucs -4.5 and has been bet down to -4 at some spots with a 55% tickets coming in on the Saints. I think the fair value is Bucs -4, which is where the market is going. Since the contest is offering the hook, I think the Saints are the value play here—though hardly the sharp angle.
Unofficial Plays: