It’s wild we’re already at the final week of the fantasy regular season. There will be plenty of time for postmortems, though we did already start some of that discussion this week on Stealing Bananas, in what was one of my favorite episodes of the season.
For the intro today I want to talk about fantasy tools, and how we analyze decisions that are in many ways out of our control. I’ve talked before about this illusion of control, and about how when you scale to the best high-stakes DFS players playing in smaller-field contests, that these people are so indifferent to the narratives around matchups and situations — at least beyond the extent that those things impact projections, i.e. they aren’t thumbing the scale and double-counting — that they are often far more focused on the ownership projections and how to play their opponent (i.e. what other people in the contest might do). The result can become leaning into a “tough” matchup, or bad weather, or something else that feels like it isn’t a great situation, but still carries enough upside to lean into, when the payoff is disproportionate due to those ownership effects.
Typical start/sit decisions aren’t the same, but there are lessons to unpack from these players and processes. One of them is this idea I’ve been toying with recently that most tools in the fantasy space are designed to make better fantasy content, rather than make better and more predictive decisions. Now that sounds pretty cynical, and it’s definitely an oversimplification where there’s a lot of overlap. But in an industry where hundreds if not thousands are competing to find the next trend, there exists a constant lack of understanding of what the data is even saying. In many cases, that lack of understanding extends to the metrics themselves, and what they calculate.
But the best tools around the fantasy space simplify these complex elements. And that’s important! There’s so much information to be researched and contemplated that paralysis by analysis is a real fear. And yet, I’m increasingly seeing data used merely to validate a gut decision. Or, perhaps a step further, used to accept or refute an initial hypothesis, but then where that first drawn conclusion gets deeply engrained. More information can’t move the individual off it.
We’ve all been down the rabbit hole with fantasy decisions, so it makes sense as a sort of defense mechanism. Rather than getting so deeply into the weeds that we risk getting hung up on semantics, we just look for something simple. Naturally, the analysis trends toward the demand, and the explanations start to become these bite-sized nuggets. “I’m in on Player X this week because Y,” when there are a dozen more things to be considered. Depending on the level of analyst, “Y” in this case can be something very obvious, or something obviously baked into a market (like ownership in DFS, or ADP in an Underdog Battle Royale draft). Sharper analysts will acknowledge those things inherently, but may still get hung up on one data point in a sea of possible arguments. Believe me, I am the furthest from immune to this.
But I see it in others as well. Some will just give a take but won’t give the evidence or defend the reasoning, because they don’t want to get into a discussion about it (which becomes clear in the rare instances where you do get them to start peeling back the layers, and realize there aren’t many there). And ultimately these decisions do become binary. I can’t begin to count the number of times over the years I’ve had someone tell me I made a good call when my take was pretty off but I got bailed out by a touchdown.
And I think I’ve come to the idea that it gets back to the tools, and even the data, much of which isn’t very predictive. I do hate that I’ve become a “watch the games” truther, but I mean as an evidence-based analyst who has put his name behind his predictions for nearly a decade now, I will simply say that there’s no bigger edge in fantasy right now. I see the takes and the analyses on the timeline and would guess that 80% or so of analysts aren’t even watching the games on Sunday, or maybe they tune into red zone for a few hours or whatever. I used to consume football like that, and I get it, but there’s a reason I don’t anymore. There’s a ton that’s missed, and it’s all of that important nitty-gritty stuff that helps you understand the overall data. It’s the same reason highlight reels about prospects don’t tell you who they really are on a down-to-down basis.
My issue with so many of these tools and this data is most of it is helpful in the aggregate, but then it’s all baked in, and it’s ubiquitous in the analysis — everyone knows Tony Pollard’s workload has been massive but the efficiency hasn’t been there, just like everyone knows Chris Olave has seen a ton of air yards but not hit on the deep shots quite as much this year — such that the edge I referenced comes from understanding the specific situations where the data isn’t helpful. For the Olave point, we’ve in the past used the term “prayer yards” for certain players to try to delineate that while yes, tons of air yards are the goal, there are some situations where guys are racking them up with a very low chance for conversion (maybe due to QB play, scheme, what have you). That’s the stuff that when you actually watch the games becomes pretty clear.
I’ve made this case before, but the whole stats vs. film debate of a bygone era was silly, because most of the film people don’t understand fantasy football as a game, and aren’t really concerned with predicting stats. When they try to apply their “real football” analysis to fantasy, it falls not just short but way short. It misses those basic market principles I mentioned earlier, such that it’s as laughable to hardcore fantasy players as a stats nerd’s breakdown of Cover-2 defenses would be to a film guy.
When I talk about watching the games being an edge, it’s not about learning the Xs and Os and the intricacies and complexities of every scheme around the league. Ideally, we’d pick up some of that and perhaps learn to apply it where helpful. But watching the games is an edge for stats nerds to understand how the stats are compiled. It seems so obvious, but my god it’s so easy to see flawed fantasy arguments these days having just watched the games. And it’s not because I’m a meathead film bro, but because I am a stat guy and understand the trends we’re looking for and the data that matters, and then can see in real time where it’s more likely to be noisy.
And to get back to the original claim, because fantasy tools are tailored to make better fantasy content rather than better and more predictive decisions, you’re going to find ways to exploit that edge. That’s the point of an “edge,” right? That it’s exploitable. That not enough people believe it, or want to admit it. Over many years of the development of the current state of fantasy analysis, a whole lot of people who grinded every minute of film couldn’t understand data well enough to apply these things coherently to arguments about fantasy football, so the idea “I know this to be true because it’s super evident on film” is laughable.
I’ve had really a fun year this year, and my teams have had a lot of success, which is always cool. I’m obviously hopeful that will continue through the fantasy playoffs, but so much does ride on just a few weeks between now and New Year’s Eve (which, if you’re unaware, is the final Sunday of fantasy season, with no Monday Night Football scheduled for New Year’s Day).
But I also wind up exhausted by this time of year, every year. I care probably too much about the analysis itself, and about finding truth and those things. I write a ton of words, sometimes about nothing, and I’ve joked before about the irony of a column ostensibly about Signal and Noise that does wind up containing plenty of Noise. That’s the reality of writing as much as I do (this is a funny aside, but I recently wrote about how much I’ve written, and I meant to use the word “prolific” to describe my production over a certain period, but am pretty sure I wrote “prodigious,” which was a weird thing that hit me like two hours after I’d sent out the newsletter, and that is just a pretty funny typo that upgraded the meaning of that point that was funny enough to acknowledge, for any of you who caught it).
But the biggest reason I’m exhausted this year is the discussions. One thing I’ve spent a lot of time considering is if I should sort of retreat to my own corner of the internet, and get off social media like Shawn has, and just sort of focus on my own process. In many ways that’s what this newsletter is, and I’m happy to say that raising the price this year did accomplish one of the things I set out to which is that it trimmed some of the fat (with subscribers that wanted it to be something other than it is), and seemed to push my writing a little more to the periphery. But I still feel the draw of social, and I still keep an eye on what’s going on around the industry, despite it probably having an overall negative effect on both my process (as I get more reactionary) and my general sentiment about the work I do, and my contentment in it.
As I so often do, I’ve now gone done a more personal tangent than the initial point I was making. But to put a bow on it, these discussions, and the difficulty I find in having conversations that circle back to watching the games (which in terms of arguing a point just becomes “trust me,” and I hate that, and if I’m honest I would say I get sort of analytically insecure making that point) — this is all a really interesting meta about fantasy analysis and making good, predictive decisions for your rosters. Ultimately, my strong belief is everything is a case-by-case basis, and the biggest mistake is “misapplying the aggregate to the specific,” which is to say trying to round peg a square hole when a situation calls for abstract thinking.
A key I wanted to get back to was that there’s still so much unknown, and so much chaos, and we do have to respect the randomness and the variance inherent in the sport. But part of that variance I do think can be seen as the data is compiled. Really hardcore data people want you to be able to explain why you don’t trust a given stat in a given situation, and that’s where I’m getting into these personal anecdotes (because that is very difficult to do, particularly on those terms, when those people won’t just accept the “trust me” argument and want harder evidence, so it has to become about the stat as a whole, and why it does work but maybe there are specific situations where it doesn’t, and my god I’m getting exhausted just describing the type of argument I have to make).
What I’m arguing for is the selective use of data, which is obviously a no-no in almost any walk of life. But if we think about concepts like tacit knowledge, we have to be humble enough to accept that “the meatheads” as I so endearingly coined them earlier might have a reason for getting so hot about some of their arguments. Generally speaking, we can’t know what we don’t know, but we should be open to learning. A big element of my earlier description of analysts who cling to their first drawn conclusion of any research is an inability or unwillingness to have an open mind, learn, and accept differing explanations. A lack of curiosity. And because of how difficult football analysis is, and how humbling, and frankly because of how little there is to actually figure out, I do think the curious among us get weeded out quicker, while the uncurious find it fertile ground for their approach.
But I write all of this today because we’re onto the final week of the fantasy regular season. Many of you are playing for your playoff lives this week; others have clinched and that do-or-die mentality starts up next week (or in two if you secured the bye). Regardless, start/sit decisions get massive now, from here on out. And when you labor over one and get it wrong, and you lose a game that ends your whole season — which you’ve put hundreds of hours into — that absolutely sucks. My wish for all of you is that if you’re not going to win the title, that you lose your playoff game by more than any realistic start/sit decision could have decided.
But if it does come down to that, please heed the point of this intro. Don’t go seeking the analysis you should have considered.
“I should have just started my studs.”
“I should have just trusted the projections.”
“I should have paid more attention to the matchup.”
“I should have listened to that specific analyst.”
“I should have defaulted to the WR over the RB.”
I could do this all day. And most importantly, I could write every single one of those the other way.
“I shouldn’t have just started my studs.”
“I shouldn’t have just trusted the projections.”
“I shouldn’t have paid so much attention to the matchup.”
“I shouldn’t have listened to that specific analyst.”
“I shouldn’t have defaulted to the WR over the RB.”
It’s all bullshit. The analysis beforehand — as I tried to argue above, a lot of that is bullshit, too. But definitely the hindsight analysis is all bullshit.
So as you go to make your decisions, consider them, labor over them — some of them will be true coinflips — but then make them however you make them (flip a coin if you have to), and then accept that the outcome is what it is. We have such a negativity bias to the bad outcomes on these, but missing a start/sit in a key spot in the playoffs is about the same level of variance as a key injury derailing your season. It’s way more personal, and I can still remember the specifics of some of my worst start/sit calls that cost me championships, but not the ones that won me them, despite vaguely knowing they are there.
Definitely weigh the key factors, but this whole intro is also why I typically just trust the projections. When I don’t, it’s for these “input volatility” reasons, which really just means I’m saying I think I know a way the game is going to go that the projections can’t account for. And having the boldness to do that is crucial, too. Just absolutely do not beat yourself up over making a call and getting it wrong. That’s what this hobby is, and the times you get them right will feel that much better because of the frustration of the misses (as I said, I don’t remember the specifics of my great calls necessarily, but I promise I remember the emotional high of things coming together for a fantasy championship).
Alright, enough of yet another long-form concept in December when I’m freakin’ incapable of making a coherent point. I spent two hours writing this thing in the dark this morning; I haven’t seen the sun in 84 years; you can’t expect me to be articulate.
Below will be my input volatility quick notes. Even if you thought the above was some amazing discussion, the obvious takeaway would be that I am not the answer and you should not confuse my ability to argue abstract points with being better at getting all the specific calls right (just because I understand we’re all fucked does not mean I’m less fucked).
I always note that I do this piece quickly, and to defer to good projections over my thoughts, because I’m talking through ranges of outcomes, not what I believe to be explicitly great predictions. I want you to make your own decisions, and play what you believe to be the right calls, but I’m happy to write up some thoughts for consideration. Let’s get into it.
Buccaneers at Falcons
The Bucs are what they are, but there is of course no one more volatile than Arthur Smith. I don’t envy anyone trying to project this dude, and divvy up usage for what he might do in a game. I just hope they lose, because we’re firmly on a path to the Falcons making the playoffs and Smith being around again in 2024.
Rams at Ravens
There’s weather concern here, but it seems to be lightening up at least some. Certainly something to track into Sunday morning, although most of the decisions on LA’s side are sort of obvious (including Cooper Kupp, especially with Tyler Higbee out), unless you have great alternatives. (I do always say things like that and then get reminded some people have rosters where, yes, I might consider sitting someone like Kupp.)
There’s plenty of Keaton Mitchell and Gus Edwards uncertainty, but I like leaning into the explosive rookie after he built up his workload before the bye, including a start in Week 12.
I’m pretty indifferent on Isaiah Likely and would probably rank him about where most would, which I’m guessing is somewhere in the mid- to upper-TE2 range. He’s certainly a bet you can make where you need some TE upside, but also isn’t one I’m jamming in where I like an alternative.
Lions at Bears
Another potential weather game. I’m hopeful we get more Jahmyr Gibbs, but it’s a revenge game for David Montgomery and the Lions might be up enough to just roll with Monty and let him get his numbers against his old team.
I don’t anticipate any clarity on the Bears’ backfield before the end of the fantasy season, as I think all three have done enough that the team has to like what they have, and the discussions of a hot hand could lead to anything in any given week. There will be some points here at some point, but also some low-scoring weeks where certain guys take a back seat. I’m not sure I could start any of them for the rest of the year.
Colts at Bengals
Nothing really on the Colts. The Josh Downs stuff last week wasn’t great, and Zack Moss is obviously playing a ton.
I’ve written about my Jake Browning skepticism, but this is a nice matchup. He does probably project better than he should given his league-low 4.5 aDOT last week, because those numbers were just so good. The way I’d frame it, I don’t anticipate his Week 13 line is really even in his Week 14 range of outcomes — I think it’s a virtual lock he’ll be defended differently and forced to push the ball downfield more, and he’ll need to consistently succeed in different ways to reach the 350 yards he did last week. I suppose that’s possible and thus in his range of outcomes, but I’d peg it as far less likely than most, even as I did find things in his downfield passing last week to be optimistic about (here’s a bad comparison: I found things to like about Will Levis’ downfield passing right away, but that hasn’t consistently translated to 300-yard passing days; it’s bad because Browning’s accuracy vs. Levis’ arm strength make them very different passers, and also they have very different supporting casts, but the point is how flashes of success don’t necessarily equal full-game lines and there’s still a lot of risk with these young guys; you’ll recall that in his first game, his long pass to Ja’Marr Chase was caught off a deflection, and also that the Bengals have hit for at least a 28-yard screen play to Joe Mixon in both of Browning’s starts, which won’t happen every week).