This is the 174th installment of The Labyrinthian, a series dedicated to exploring random fields of knowledge to give you unordinary theoretical, philosophical, strategic, and/or often rambling guidance on daily fantasy sports. Consult the introductory piece to the series for further explanation.

I recently wrote a piece about why NFL matchups matter much less than you think in DFS. Technically, I probably should’ve titled the article something like “Why NFL Matchups Matter Much Less Than Most DFS Players Think” — because I don’t know what you actually think — but whatever: #ClickBaitLife, amirite?

I have a few follow-up thoughts on why player- and/or position-specific matchups don’t matter (much). When you have an idea with legs, you treat it like a goat: Milk it.

DFS Is a Three-Dimensional Market

Before I get into matchup talk, I think I should elucidate the way in which I view DFS. What I discuss later builds upon it. DFS is a marketplace, and it has three constitutive ‘submarkets.’ The best DFS players are good at understanding the ways in which those submarkets connect and the varying ways in which each is important to the total market. For any given professional athlete, here are the three submarkets in which he (as a DFS asset) is circulated.

  1. Player production
  2. Platform salary
  3. Ownership rate

The production market is the most basic. When you invest in a player, you’re hoping to get a lot of production out of him as measured by raw fantasy points. The secondary market is the one created through the salaries the platforms assign to players. When you invest, you’re hoping for more than just a lot of fantasy points: You’re hoping for lots of production relative to a player’s salary. At FantasyLabs, we measure the relationship between production and salary with our Plus/Minus metric. Finally, the tertiary market is created by DFS participants in their contests, be they cash games or guaranteed prize pools. When you invest, you’re hoping that a player can have great salary-based production relative to his ownership rate. At Labs, we project ownership rates in our Models, and we also have historical ownership rates available in our Ownership Dashboard, Contest Dashboard, and Trends Tool, which are perhaps the most valuable tools in the industry.

With these three submarkets in mind, the best DFS asset ever would be a player who leads the slate in scoring at minimum salary and almost no ownership.

Many DFS players know that raw production, salary, and ownership are correlated, but they haven’t thought about the interconnectedness of these factors as much as they should. It’s rare for someone to be good at DFS and not understand the three-dimensionality of the overall market.

I mention all of this because it has a world of bearing on how I think about matchups.

Matchups Don’t Matter — Except That They Do

I’ve written before about the double-counting error. In DFS, I think there’s a double-counting error — maybe even a triple-counting error — when it comes to matchups.

In my first piece on matchups, I focused mostly on the primary DFS submarket of player production. In a predictive sense, my working theory is that for many players, especially running backs and wide receivers, matchups don’t matter much. We should be aware of them — I take them into account when making my FantasyPros rankings — but often a team’s identity matters much more than a team’s matchup. If an offense thinks of itself as a running unit, it will often be inclined to run the ball even when facing good rush defenses. If someone is an addict, the last person to know is usually the addict. Matchups matter little because few NFL teams substantially alter the way they play based on their opponents.

And yet despite what I just said, in a pragmatic sense matchups matter a lot — because many projectors, rankers, and DFS players probably overvalue them when thinking about how many points players are likely to score in a given game. What I’m talking about here is the difference between important and predictive variables. Matchups aren’t hugely predictive, but they’re important because the market thinks they’re important.

The Multiple Accounting Error

And here’s where we get to the error of multiple accounting: Matchups aren’t overvalued just in the player production submarket. They’re also probably overvalued in the platform salary submarket and definitely overvalued in the ownership submarket. Matchups are given too much predictive weight across the entirety of the industry, especially if we take into account how the salary and ownership submarkets work together.

If you do even a little research, you’ll notice that in the aggregate salary and ownership tend to correlate. There are exceptions to this — when a starting running back suffers an injury in the middle of the week and the backup is cheap, he’ll probably be chalky — but in general the players who are more expensive tend to have more exposure. That’s natural. The issue is that the platforms take matchups into account when setting salaries. The sites, especially DraftKings, have been highly responsive this year to perceived weekly changes in situations, including matchups.

Not many people think of DFS salary as the equivalent of the spread in sports betting, but they should. Just as bookmakers set lines, DFS sites create salaries that are intended to take matchups into account. Whether they do that efficiently is another matter: The point is that adjustments are made based on opponents. Salaries are matchup-dependent player-specific spreads. Matchups are taken into account in the salary submarket.

And matchups definitely come into play in the ownership submarket. If you see that a team has a high implied Vegas total (and totals are tied to matchups), you’ll be likelier to include players from that team in your lineups. One of the most contrarian things you can do is ignore or fade Vegas totals, because so many people anchor to the lines when analyzing slates. Fading implied expectations makes even more sense when you remember that the matchup you’re looking to exploit via the Vegas lines has already been taken into account in point projections and salary adjustments.

I Guess Matchups Do Matter

In Shakespeare’s Troilus and Cressida — one of his worst plays and memorable for only a few passages — Hector argues with his family that Helen of Troy is not worthy of a war: “Brother, she is not worth what she doth cost.” In his speech, Hector suggests that value is intrinsically derived. What matters most is not what someone is willing to pay but what something is actually worth based on its own merits: “But value dwells not in particular will.”

In DFS, the value of matchups far too often dwells in particular will. Matchups matter to the extent that people are misestimating their worth. Otherwise, they are worth little.

What all of this means is that it’s easy for DFS players to overvalue matchups in the aggregate. If a wide receiver has a great matchup, the average DFS player will inflate his point projection, overlook the extent to which his salary has been adjusted, and then roster him in too many lineups. Through this negative DFS triangulation, people are basically overpaying for matchups.

Matchups matter — primarily because most DFS players don’t appreciate how much they don’t matter.


The Labyrinthian: 2017.79, 174

Matthew Freedman is the Editor-in-Chief of FantasyLabs. He has a dog and sometimes a British accent. In Cedar Rapids, Iowa, he’s known only as The Labyrinthian. Previous installments of the series can be accessed via the series archive.