LESSONS LEARNED FROM THE 2022 DFS SEASON
The dynamic game of Daily Fantasy Sports (DFS) requires much more than simply knowing the sport for which we’re entering contests to be successful. We must be adaptable, precise, and open to learning from previous endeavors, the latter of which will be the primary focus of this weekly written piece. Game Theoretic methodologies will allow us to analyze and dissect the previous week’s winner of the largest and most prestigious Guaranteed Prize Pool (GPP) tournament on DraftKings – the Millionaire Maker. These same tenets of Game Theory, which can most simply be explained as the development of decision-making processes given our own skill and knowledge, assumptions of the field based on the cumulative skill and knowledge of others playing the same game, and the rules and structure of the game itself, will allow us to further train our minds to see beyond the antiquated techniques of roster building being employed by a large portion of the field. Approaching improvement through these methods will give us insight into the anatomy of successful rosters and will help us develop repeatably profitable habit patterns for the coming weeks.
We’re going to continue our macro journey and examine some of the things we’ve learned this season, reflect on the process of building a positive expected value (+EV) DFS roster, and apply those lessons to the short slates that the major DFS sites have provided to us for the Divisional Round.
Lessons Learned This Season, Applied to Divisional Round Weekend
Right off the rip, the dynamics of the four-game slate on DraftKings are much different this week when compared to the six-game, three-game, and two-game slates available to us last weekend. All four games carry game totals between 46.5 points and 53.0 points, with the highest game total being the first game of the weekend in Jaguars @ Chiefs. While there is no doubting that this game is the top expected game environment on the slate (this game contains two of the bottom three defenses by DVOA still remaining in the playoffs), it is likely to garner immense ownership due to it being the first game played on the weekend.
This niche aspect of the slate is attributable to human psychology, crowd psychology, and our desire for instant gratification. It’s the same reason why it is so profitable to play the full slate of games in a standard week (Thursday to Monday) and fade the Thursday game. As a general rule, ownership will be inflated in the first game of a slate that has games played sequentially instead of simultaneously.
That creates an interesting dynamic to the slate in that the remaining three games also carry high game totals, providing the potential to generate leverage through targeted exposure to the first game of the weekend. As in, while this first game carries the highest game total, it also carries the highest spread - there are significant cases to be made for an underweight stance here.
Due to the constraints placed on the field in our beautiful game, there are some instances that generate natural funnels of combinatorial ownership (I classify these elements as “restrictive chalk” and “expansive chalk” for the way they interact with the remainder of a roster). Theres also an element of positional scarcity that comes into play in this discussion, which is introduced via the state of the NFL. On this slate, and considering the previous discussion, we can be fairly certain that Travis Kelce will garner immense ownership. Furthermore, due to the lack of viable value plays available on the slate, we can be fairly certain that Kelce will be paired with Patrick Mahomes at a lower frequency than he otherwise should be.
Without going too far into the theoretics here, tight ends should be paired with their quarterback at a higher rate than the field typically utilizes due to how tight ends derive the majority of their value in today’s NFL game (touchdowns). As in, it is much more likely that a wide receiver or running back provides a GPP-worthy score on their own when compared to a tight end. Take a look back through Travis Kelce‘s game logs this season - Kelce scored more than 20 DraftKings points on nine separate occasions this season. He did so in Week 1, Week 4, Week 5, Week 6, Week 9, Week 10, Week 11, Week 15, and Week 16. Patrick Mahomes averaged 31.79 DraftKings points per game in those nine contests. Patrick Mahomes averaged 25.39 DraftKings points per game in the eight games in which Travis Kelce failed to crack 20 fantasy points.
But if Patrick Mahomes theoretically should always be played on Travis Kelce rosters, that generates an innate roster funnel due to the high combined salary required to play both pieces, which leaves less salary per remaining roster position available for the rest of the roster.
To bring that discussion full circle, highly owned expensive pieces naturally force rosters to be built in a similar way, which leads to another concept that I created to help visualize the interaction of agents through the confines of a salary cap game - the chalk build.
The chalk build (how individual agents interact through the confines of a salary cap game) requires multiple inputs in order to return actionable information, including expected ownership, the availability of viable salary-saving value options, slate dynamics, perceived certainty, and more. We’ll be examining this principle prior to the release of expected ownership, which makes it more a theoretical discussion than a discussion on how the theory pertains to this individual slate.
As things currently stand, there isn’t a ton of viable value options available to us on this slate. Combine that with the natural roster funnels discussed in the previous section (in addition to the state of the quarterback position on this slate) and we’re left with a slate where more balanced builds are likely to completely dominate the entries in play. This makes sense logically as the field is highly likely to prioritize perceived certainty at the quarterback, running back, and tight end positions - running back is the most projectable position in NFL DFS while quarterback and tight end are usually positions of “haves and have-nots.”
The theoretical discussion gives us a good idea of easy ways to generate leverage on the four-game Divisional Round slate. Multiple pay-up running back builds, multiple pay-down running back builds, multiple pay-up wide receiver builds, and Patrick Mahomes plus Travis Kelce are a few examples of how to smartly leverage that understanding.
As alluded to in the opening section, all four games this weekend carry inflated game totals compared to what we’ve grown accustomed to this season in a down-scoring year around the NFL. Looking at some of the top-level metrics used to analyze game environments, we find that six of the remaining eight teams are the top six teams in the league this season in overall DVOA (all but the Jaguars and Giants), five of the remaining eight teams are in the top half of the league in defensive DVOA (all but the Chiefs, Jaguars, and Giants), and all eight remaining teams are in the top half of the league in offensive DVOA. The important thing to realize through this discussion is that a strength-on-strength matchup, as is the case for a game like Bengals @ Bills, yields a wide range of potential outcomes as for as the potential game environment goes.
So, while the top expected game environment is very clearly the Jaguars @ Chiefs, any one of these other games could provide the top actual game environment on the slate due to how good each respective offense is. Ranked in order of likelihood to produce a top-level game environment, I would put them in the following order - Jaguars @ Chiefs, Bengals @ Bills, Cowboys @ 49ers, and then Giants @ Eagles.
Range of Outcomes vs. Median Projections
There is a general misunderstanding with how projections systems develop their projections around the industry, primarily due to the need to provide information in easily digestible ways for mass consumption. That has led to the primary projections most of the field utilizes when building their rosters being median projections, which is the nice single-point value you see when you pull up projections from most top sites around the industry. The definition of median projection means that the eventual outcome will outperform and underperform the value shown at an equal percentage if you were able to play out each individual slate an infinite number of times. These projections systems are continually back tested to refine their processes, which helps to improve their accuracy for future slates.
That said, simply understanding that projections seen from major sites are median projections can help us visualize how to leverage that information. Some sites also provide range of outcomes projections, which are much more useful in determining the viability of an individual play. To help visualize this process, think about a standard bell curve plot - with various data points represented through both frequency and standard deviation. Some players possess a tighter range of outcomes, which would make the bell curve appear more tightly grouped around the median outcome as well as higher in magnitude. Some players possess a wider range of outcomes, which would make the bell curve appear flatter and more spread out.
Most fantasy players understand that general concept, which is made easy by standard industry connotations like “safe, GPP-only, boom-bust, and cash game play,” but the general understanding of what is going on behind the scenes is more or less lost in translation for most of the field. Understanding this concept and being able to apply it to an entire roster (and beyond just individual plays) will drastically improve your fantasy play!