Week 8 Reflections: Market Inefficiency?
Welcome to Throw The Dam Ball!
If this is your first time reflecting, I'd recommend first reading the inaugural piece of this series to get a sense of what I'm trying to do in this space.
This week we cover(and if you want to skip to those parts): General Market Trends, Team Trends, My bets, and my DFS lineup.
General Thoughts and Trends:
If you have been following @sportfoliokings, the good guys there have been beating the drum that favorites are winning at an unprecedented and likely unsustainable rate. It was only a matter of time until underdogs started winning some games, and this week we saw some of that regression. I’m looking at: AZ, CLE, CIN, MIN--and to some extent TB & LAC--- all being beat this week as big to moderate favorites.
I was on a couple of these underdogs leading to my best week relative to the market in some time. Keeping with this trend from this week, I would expect more upsets going forward. Anyway, I’m currently sitting in 196th place out 864 in the @leesharpe prediction game.
We often talk about EPA/play in more analytical circles, but we very rarely talk about EPA per drive. But NFL offenses scheme and design plays sequentially-- viewing the operative units as drives, not plays. I think this lens is perhaps a more valuable way to evaluate teams play thus far. Here is the EPA/drive on both offense and defense for each team in the NFL through 8 weeks.
CLE: I think it might be time to adjust my prior on CLE. I came into the season thinking they had a shot to compete for a SB. But after 8 weeks, I think they’re probably closer to a fringe playoff team. Baker finished the 2020 season really strong(PFF’s 8th best QB by season's end), but has taken a big step back this year: 24/34 in PFF grade, 19th in EPA/CPOE composite. Injury or not, and despite the strong run game, unless Baker regains his form from last year, the Browns are about a league average offense—and the data above seems to support that.
MIN: I was quite surprised to see MIN so far to the left on this graph. Kirk Cousins is PFF’s 2nd ranked QB! But this team has been absolutely dreadful running the ball, with a Success Rate of only 30%. And despite this inefficiency, MIN throws well below expectation, with the 8th lowest Pass Rate Over Expected(PROE) at -2.5%. Maybe it is time to start trusting Cousins and letting him throw the ball.
AZ: AZ has been the best team in the league so far by EPA standards. And this defense, as you see, has played really really well even if they are (rightfully) overshadowed by how good this offense has been. Despite being 1st in both EPA categories, most power rankings(538, FO, PFF, FPI) have them closer to the 5 spot than the 1 spot. Will be interesting to monitor how the market prices them going forward.
DET: The Lions are so so bad. Early in the season it looked as though they played hard and could at least compete in some games and cover spreads. But they basically break my graph here-- they are so so bad.
TEN: The Derrick Henry injury sucks, full stop. Looking ahead, my best guess is that D-coordinators will stop loading the box and Tanehill will struggle a bit more as a passer with more guys in coverage. I’m ready for the Henry truthers to come out saying: “see, Henry was the engine of the offense.” hint: I don’t think that is true. But let’s move on to this week’s bets.
Bets:
IND Div+400 & TEN +130
I’m working through what I think is a slight market inefficiency between futures and game-by-game markets. A word of caution: this is only a ½ baked idea at this point and I have not fully run the numbers.
The idea is to use the IND division as a proxy for betting IND this week. I am basically arguing that short and long term markets are not perfectly efficient and manipulating different points in time in each market could be a way to profit. I will go through the bet and then my conceptual understanding of how this inefficiency could exist.
The situation: IND’s schedule week 8 and 9: TEN(-3) & NYJ(-14). TEN week 8 and 9: @ IND(+3) and @ LAR(+4.5). My bet was that if IND goes 2-0 and TEN 0-2, I would be able to grab TEN div at + odds and have both sides of the AFC South, guaranteeing a profit on the initial bet. And what’s more, because IND was favored to win this week, I was able to hedge with TEN at + money to add another layer of potential profit. I was basically guaranteeing myself a profit with every outcome besides a combined Jets and Rams win in week 9, which I, and the markets, thought was highly unlikely.
In stages with some arbitrary numbers :
Colts to win AFC south $10->$40 & Ten wins+130: $10
If TEN wins: (30% profit)+less valuable, but not yet dead, IND Div ticket
If IND wins…..
Assume LAR+NYJ win as heavy favorites
Grab Ten Division before Week 10 at + odds: $20—> profit!
The way I understand this bet theoretically: This bet basically manipulates the futures markets models at different points in time. It is like financial volatility. In essence, odds are fluid and it all depends on your time of reference. Imagine a simulation that incorporates all available information and goes through a straight line with 10 points but only 2 visible points: A & Z. A represents where we are now-- right before week 8. Z represents after week 18, or the end of season. The +400 division betting odds corresponding to this “simulation” only visibly exist in these 2 points in time. In other words, we as bettors can only see A and Z. But there are also 10 points(the weeks between Week 8 and Week 18) where there are fluctuations in our initial line. If we arbitrarily stopped at, say, point C corresponding to Week 10 in the simulation, we might find at that point in time the odds are not actually +400(20% probability, but are actually +150(40% probability). And then, as we continue this simulation through Week 18, we see that the odds at point Z will be +400/20%. Basically, we can leverage or predict volatility since those middle 10 points are hidden in the final price we see now, though they are not absent from the simulation.
More proof for potential market inefficiency: This game opened as a PK and TEN ML was -110. It worked its way up to TEN +150 at one point, but the division market’s odds did not budge.
Either way, this bet netted me a nice profit since TEN won. And the Colts ticket is not yet dead, with Henry out for the year, and the soft schedule for IND over next few weeks. Maybe I will still be able to grab a TEN + division bet at some point in the next few weeks!
SF -9.5@ CHI (+190)
This article is already running real long, yikes. I’ll try and keep these short.
SF is in a different tier than CHI. If you take a look at the Drive EPA numbers above, you can start to get a sense of the difference in caliber of these teams. Do I think SF is a Super Bowl team? No. But can they compete for a playoff spot, certainly! And that is far different than CHI who is a bottom cellar team unless Fields improves drastically.
Why the Alt line? The CHI offense is really bad, full stop. And very bad teams are recipes for blowouts(all 5 CHI losses now have been by +10 pts). And I think what separates a team like SF from LAR let say, is not the ability to score 30 points-- both teams can do it. But the difference lies with whether each team needs a particular matchup to hit that ceiling or whether teams score matchup independent. LAR scores points every week, SF does it only in certain spots.
Well, against a bad Bears D missing Khalil Mack, this was one of those spots for SF to hit their offensive ceiling. And with CHI’s anemic offense, I did not think CHI stood a chance to keep pace offensively and this would end up a 2 score game.
Besides, it has been a trend for some time, that SF has been able to protect and extend leads well. While not the biggest of samples, you can see below SF has done a great job building and extending their leads. This gave me the extra boost that should SF jump out to an early lead—as I thought they might— SF would not let CHI back into the game.
JAX -5.5+300
We noted a few weeks back that Trevor Lawrence has begun playing better. It should not surprise us to see a rookie QB improve as the season rolls along. In short, if I would project T-Law>Geno from this point in time onwards, and I like betting on spots with a QB edge.
But this bet was mostly about understanding the difference between how teams should attack vs how they are attacking. The Seahawks, despite facing teams stronger defensively vs run, have been running 8.7% above expectation since Russ went down. This fits with the well established narrative of the Seahawks wanting to run the ball. It would be a leap of faith to say that the team will finally start passing this week. And of course, considering how porous the JAX defense has played--last in DVOA, 29th yds per attempt(I can cite so many stats here). But by insisting on running the Seahawks would be losing their biggest advantage in this spot.
Why 5.5? I saw this as a spot where the Seahawks could easily fall behind early by insisting on running the ball despite little success, and with T-Law carving up this Hawks secondary early. In this game script, I thought there was a decent shot the Jags would jump to a lead and be able to hold a one score lead through the back door.
Reflection: If I liked that angle, and it was the driving force behind my bet, I should have bet an alt 1st half line. I’m chalking that up as a mistake, though I don’t hate the overall process of JAX>SEA this week.
MIA O 21 points (+200)
First things first, and what is at the heart of this bet: 21 points is not many points to score, even if the Bills defense has been good by every metric:
Why this spot? Both teams play with extreme pace and barely run the ball giving this game elevated pace and pass attempts. When we can project adequate QB’s to drop back to throw 40+ times I like their chances to score 3+Td’s on volume alone. (Obviously there is more context involved—I am not suggesting to blindly bet on this trend.) But with Parker back, the recent trajectory of this offense, and volume on their side, I liked my +200 odds on Dolphins to score only 21 points.
4 team ML Parlay: KC+SF+NE(+410)
Why the Parlay? Well I didn’t really like the tease in this spot but I liked these games as betting opportunities. Why didn’t I like the tease? I never ever tease through zero(SF was -4), and KC only wins one score games, it seems so I did not want to be on the wrong side of 3 as they were 10.5 pt favorites.
KC: Until something changes with KC, I will save my explanation.
SF: see above!
NE:
Last week I highlighted why I am higher on the Pats than most and why they are trending upwards. This team has had some unfortunate turnovers through 7 weeks and, as you see below, they’ve lost close to 60 expected points on turnovers alone. I see those turnovers as more noise than signal. This team is legit good, and when the turnovers regress, the team’s play and analytical profile will only improve.
I also think this was a perfect matchup for the NE defense to slow down LAC. Why? Belichick for years has done a great job shutting down teams top options( or 2 when he has the personnel for it). The Chargers are extremely top heavy on offense and without any ancillary receiving threat, I thought they might struggle to move the ball.
Bottom line: while I love Staley and the Chargers and am bullish on them long term, I think this game was closer to a toss-up, and the odds did not reflect that. I want to start taking some calculated shots to really build up the bankroll, and I was confident in KC and SF's ability to win, giving me a nice return on this parlay. Upon reading this over, I liked my process for my bets this week and it worked out quite well this week for the bottom line :)
DFS:
Another extremely mixed bag for me this week. I think I identified the right side of the right game environment. But I am a little upset with the Callaway play, and that I overreacted to late news forcing in Pitts and Swift right before lock. Let’s break it down:
I thought the NO vs TB was the best game environment on the slate. We have 2 teams who were going to pass a lot. TB has the highest Pass Rate Above Expected (PROE) and TB forces opponents into the highest PROE. So we knew the passing volume would pile up here.
The question was which side I wanted to take. I knew I wanted Godwin and Tyler Johnson. I don't usually buy into WR-CB matchups, but I was willing to make an exception for Evans-Lattimore given the history that 1) Lattimore will actually shadow Evans 2) Lattimore has consistently slowed Evans production. This meant the targets would filter to Godwin and Tyler Johnson. Godwin is an elite receiver and with talent+role+price I was going to make him a priority play. The same can be said for Tyler Johnson who saw the field 80% of the time with AB out, and was only priced at 3.1K, right above the 3K minimum.
And of course Brady would have made sense as my QB, but at his price point 300+yards+3 td’s would be a solid but not elite price considered score. And besides, with no rushing ability, I thought I could capture most of Brady’s upside in those WR’s. I also liked this spot for NO(see prediction game) and thought the Saints would be able to hang with the Bucs, and maybe even pull the game out. Considering you don’t run against the Bucs, this means the production would have to flow through Winston.
At 1.4K cheaper, Winston was priced for a guy who has yet to really throw the ball this season. But as we discussed, he would be forced to throw in this spot, giving him some really good upside for his price. Also, Winston has been rushing the ball of late, which added about 3-4 points to his floor. If I liked NO to keep it close and maybe even win, and they were not going to be able to run, then I had to think Winston would at least go neck-neck with Brady. With rushing ability+price discount, I thought this was a sneaky great spot for Winston. And Winston was proving the thesis perfectly, until he went down in the 1st half. Yet again, we had a crushing DFS loss to injury. Absolutely brutal.
But I didn’t stop there attacking this game, going to Callaway in my Flex with Winston. My thought process was that Callaway had over past 2 weeks had emerged as the #1 receiver running routes on basically every snap and generating 23% target share over past couple weeks. With his deep threat ability and big role, I thought he was a sneaky stack partner with Winston.
Reflection: This was bad process. I let the short term noise/usage dictate this play rather than years of evidence that the Saints spread the ball more than any other team and snap shares and targets are not consistent week-to-week. And, not to mention, Deonte Harris and Trequan Smith were either hurt or being eased back in the past couple weeks, and there was no real reason to think Callaway would retain his usage with those guys back in the fold this week.
Henderson: We’ve been through this drill. Henderson sees elite usage, attached to one of the NFL’s best offenses as a massive favorite. Not overthinking this one.
AJ Brown is a beast. And he was set to have a huge target share in a great game environment— sign me up!
Ok we’re going too long so I won’t break it down fully, but I loved this game IND/TEN game environment. I debated a Wentz lineup but I thought I could capture upside through a parlay with: Wentz TD’s+Completions+Pittman and Brown O Rec yds. Lo and behold we are up 8 units on that player prop parlay. Maybe I will re-visit this next week and explore how this angle supplements my DFS and bets in my “football profit process”
Back to DFS: to round out the lineup we had Swift and Pitts. Neither of these guys were on my radar until Sunday noon. Why? I did not like the game environments in each of these spots. For DET @ PHI, I thought both teams would try to run the ball, enjoy low efficiency leading to a game with few possessions and points. As for CAR vs ATL: CAR has quietly put together a great year defensively(drive EPA chart above), and well, their offense has been anemic. ATL’s offense has struggled this year as well. I did not think points would pile up in this spot. Given I did not see this game supporting multiple “A level” scores, it would be a guessing game who would be the ones to score some fantasy points.
But as the J- Williams inactive news came out, I thought I had to jam in Swift, and with the savings from Ekeler down to Swift I was able to move up from Goedert to Pitts. But I hate reacting to late news. Why? It often leads to decisions determined by short-term recency and availability bias, and a “fear of losing” approach rather than sticking to what the week’s research led me to. Don’t make decisions last minute unless something really is a can’t miss change ie; a 4k running back with a 3 down role becoming available.
Results:
Took an L this week, though we were 6 points out from cashing and doubling our money. Of course, if Winston played the full game I would have cashed. *insert Raymond Holt GIF: Paaaaaaaaaainn.
That’s all for this week!
Thanks for reading my adventurous thoughts and I look forward to being back in this space next week! We’re so blessed to be able to watch and talk about football, aren’t we?
You can follow my work: @throwthedamball. My DM’s are open — I really love all kinds of suggestions, comments, criticism!