As the “DKLeak” scandal continues to develop and garner ever-greater attention from the mainstream media, the focus has tended to be on the legal aspects of daily fantasy sports (DFS), how and when the leak itself happened, and what the various parties involved are saying and doing about the game integrity issues which have come to light.
At the same time, reading comments from readers and casual DFS players around the web, what I’ve been seeing is that many people are not quite clear on the specific nature of the problem. Almost everyone intuitively understands in general terms that it’s a potential problem when an employee of one site, with access to internal data, is allowed to play on another. When it comes to the specific data which was leaked, however – that is, player ownership percentages – many people don’t seem to understand why that information in particular is a big deal. That, in turn, stems from a fundamental misunderstanding about the strategy of daily fantasy sports, particularly when it comes to guaranteed prize pool (GPP) contests.
What is “ownership percentage”?
At this point, the scandal has been garnering so much attention that it’s made the radar of many people who have perhaps only heard of DFS in passing and not tried it themselves. So, to start at the very beginning, allow me to provide a nutshell explanation of how DFS works and what data was leaked. I’ll use the term “contestants” to refer to the users of the site, to distinguish them from “players,” by which I mean the athletes competing in the sport being wagered on.
A DFS contest takes place in a single day, involving all the matches played in a given sports league on that day. Contestants draft a team of players who will be competing on the day in question; they’re constrained in their choices both in terms of the number of players they can draft, and a “salary cap,” where salary is an abstract points system meant to reflect the expected quality of a given player. Following the real-life competitions, the players are assigned a point value based on their performance, and prizes are then awarded to the contestants with the greatest total points for their drafted players.
Ownership percentage, then, is a player-by-player statistic reflecting the percentage of competitors who selected the player in question for their team. This can be as much as 25-50% or more for some star players, down to a few percent, or even a fraction of a percent for players seen as having poor value.
So is ownership percentage like “polling the audience”?
The biggest misconception I’ve seen among lay people discussing the scandal is that the issue involves insiders using the supposed wisdom of the crowd to choose their lineups – that is, basing their picks on who other contestants believe is good value. This then leads to the argument that it shouldn’t be such a big deal, because winning at a raked game like DFS or poker requires one to be better than the average, so knowing how the crowd is betting shouldn’t be much of an advantage at all.
The reality is quite the opposite: The actual advantage in knowing about large-scale ownership trends is in being able to selectively bet against the crowd. This error in the public’s understanding of the situation stems from a fundamental, but natural misunderstanding of the game’s objective.
It’s not about maximizing average score
On the surface, the basic objective of DFS seems obvious: to maximize one’s score. DFS is a high-variance game however, as it’s hard to predict how an individual athlete will perform on any given day. Thus, when we talk about maximizing score, we’re really talking about maximizing some kind of average, such as mean or median expected score.
The thing to realize, however, is that a contestant’s absolute score is irrelevant and it is rather their relative score which is important. The actual objective of the game is not to score as highly as possible, but to have a score higher than that of as many opponents as possible.
At first glance, these two objectives seem identical; after all, how could a lower score beat more opponents than a higher one? They would in fact be identical if every contestant’s score was based on independent factors, but this is not the case. Contestants’ scores are all based on the performance of the same pool of players, and the performance of any given player affects a contestant’s score only relative to other contestants who did not draft the player in question. Therefore, the performance of low-ownership players on one’s team has vastly more impact on one’s overall position in the standings than that of one’s high-ownership players.
To see this, consider the extreme example of a player who has an ownership percentage of 100% – that is, who has been selected by every single contestant. That player’s performance is in fact completely irrelevant to the final outcome, because it affects everyone equally. Whether he has a breakout performance or is taken out by an injury on the first play of the game, none of the DFS contestants gains or loses any points relative to anyone else.
Looking at things from the relative perspective, if you hold a player with 25% ownership, his performance will gain or lose you ground relative to 75% of your opponents, but he will always be a break-even prospect for you relative to the other 25% who’ve chosen him. A player with 10% ownership, on the other hand, can gain or lose you ground against 90% of the field. Similarly, the decision not to draft a high-ownership player can have as much impact on one’s relative outcomes as the decision to draft a low-ownership one; in the case of a player with more than 50% ownership, the player’s performance affects those who didn’t draft him to a greater (and opposite) extent than it affects those who did.
All of this is particularly relevant in GPP tournaments of the sort which produced the $350,000 win for DraftKings employee Ethan Haskell, which in turn caused the scandal to come to light. These contests pay out the vast bulk of the prize pool to the top few contestants, so the emphasis is on maximizing the variance in one’s relative scoring prospects; having a team of players who all do pretty well is of no use if many other contestants own them as well. Pulling away from the pack requires having at least a few low-ownership players who do unexpectedly well, while hoping for weaker-than-expected performances from some of the high-ownership players that one did not take.
Of course, there’s a reason that high-ownership players are highly owned; they’re seen to be of higher-than-average value in terms of expected points versus salary cap. Thus, there’s a balancing act to be done; obviously, a contestant does want to maximize the expected points produced by his or her team, but wants to do so using as many low-ownership players as possible in order to produce separation from the pack and a chance at the top prizes.
The skill factor in DFS
This is really where a lot of the skill in DFS comes from. Just as a stock market speculator wants to identify stocks which are undervalued and buy them while they’re cheap, someone looking to profit at DFS has to predict ownership trends, and select high-potential players for his or her team during weeks when the bulk of contestants will be looking elsewhere.
Knowing ownership percentages is of course only half of the equation. It tells the contestant only which players are seen to be of high or of low potential in a given week, and the contestant must then rely on research or instinct to guess at whether the actual potential of each is higher or lower than commonly believed. A contestant without insider information, however, must guess at both these things – not only how valuable a given player actually is, but also how valuable he is seen to be. It’s clear, then, that a contestant with access to privileged information about ownership percentages holds a significant and unfair advantage over contestants who do not.
Alex Weldon (@benefactumgames) is a freelance writer, game designer and semipro poker player from Montreal, Quebec, Canada.