Ed. note: The author is the COO of Star Fantasy Leagues.
[toc]The foundation of the burgeoning fantasy sports industry in the USA is built on the premise carved out in the 2006 UIGEA that companies offering daily, weekly, or season-long fantasy sports contests are operating games of skill, and therefore are not subject to federal gambling laws.
Some lawyers, policy makers, and critics contend that one-day fantasy contests are gambling, and the question of whether these games are games dominated by skill or by chance has not been broached in a state or federal court.
While the definition of ‘gambling’ differs state by state and cannot be ignored when fantasy sport operators choose where to offer their contests, Star Fantasy Leagues has set out to prove that winners of its daily and weekly style fantasy sports contests are not simply left to chance.
If you’ve ever played fantasy sports – and I have since I was first introduced to it over 15 years ago – you already know that it takes some semblance of skill to consistently remain competitive in your fantasy leagues.
I’d think that any operator who chose to start and build a fantasy sports brand was comfortable with the fact that these games are, at the core, skill-based games.
There is an incredible amount of player data – in every sport – available for fantasy gamers to utilize in order to craft optimal teams in season long fantasy competitions, and even more data and nuances to consider when putting together lineups for daily or weekly style games.
And common sense would dictate that in order to be a successful general manager of a real sports franchise, which is what fantasy sports contests essentially allow you to do, you have to know something about what you’re doing in order to beat the people you’re competing against.
You’re reading this today because I’ve been asked by Chris Grove, the proprietor of LegalSportsReport, to write an op/ed about what Star Fantasy Leagues has recently accomplished alongside Gaming Laboratories International (GLI) and why it is important for the entire daily fantasy sports industry moving forward.
Star Fantasy Leagues commissioned GLI, an industry leader in game testing and certification services, to perform a skill simulation analysis on its fantasy sports contests.
Measuring how much skill matters in DFS
Specifically, Star Fantasy Leagues set out to see if skilled players dominated unskilled players in weekly salary cap style fantasy football contests, which many consider to be the most volatile fantasy sport game in terms of scoring and unpredictability.
Star Fantasy Leagues is currently in its second season of game testing for its fantasy basketball contests and final test results will be released after the conclusion of the 2014-2015 NBA season.
How the game works
Before I get into the football test results, here’s some context on the games SFL submitted for testing.
Daily or weekly fantasy sport contests differ from a traditional season-long fantasy sport contest in that you’re condensing an entire season into a single day to play a daily fantasy game.
Instead of drafting a team at the beginning of a sport season and managing that team until the season is over, you’re drafting and managing a team for a single day or week. The fantasy sports season starts and ends in the same day or week, and this creates a new, ever-changing game dynamic that I think is one of the best things that has been created for a sports fan in my lifetime.
The salary cap format
The weekly style fantasy football contests SFL tested with GLI are ‘salary cap’ games, meaning you, the player, must put together the best roster of players you can across a single week’s fantasy football contests, all while staying under the salary cap provided.
In short, you become the general manager of a fantasy football team for a week, then do it all over again the next week.
For example, let’s assume you have a $100,000 virtual salary cap for a fantasy football contest. At Star Fantasy Leagues:
- You must fill a team with 2 quarterbacks, 2 running backs, 2 wide receivers, a tight end, 2 FLEX spots (RB/WR/TE), and a defense – a total of 10 spots.
- Each individual player has a different salary. DeMarco Murray may cost $15,000 in cap space while Trent Richardson may cost $5,000 in cap space.
- The aim is to put together the best team you can, balancing different player salaries, in order to make the most optimal fantasy lineup and score the most fantasy points against others who are doing the same thing, all while meeting or staying under the salary cap.
- The player whose team scores the most fantasy points wins the contest.
For SFL fantasy football contests, player salaries are generated algorithmically and update each week based on the previous week’s statistics. For daily contests in other sports like baseball or basketball, player salaries update every single day based on stats from as recent as the day before, alongside other factors built in when considering salary generation.
These results of our work with GLI, while likely translatable to similar game styles offered by other fantasy sports operators, are only based on the Star Fantasy Leagues salaries and game structure.
How the simulation worked
After testing the Star Fantasy Leagues weekly fantasy football contests over the 17 week 2014-2015 NFL season, GLI simulations found that the “skilled” player beat the “unskilled” player 69.1% of the time.
It seems pretty cut and dry that this should show skill – no matter how little – dominates over pure chance.
Let’s dig deeper into it.
A skilled participant was defined as a participant with full knowledge of the fantasy performance of all players for the previous week at the moment of roster selection. The skilled participants chose their rosters based on the fantasy performance of players from the previous week and their respective weekly fantasy salary.
This did not include additional information that a human might use to craft a lineup, including injury status, strength of an individual player against an opposing team, or any other skill point. We would contend that adding these dynamics in against a purely random lineup would increase the win rate much higher than 69.1%.
An unskilled participant was defined as a participant that chose their lineups at random from the list of eligible players without any informed decisions. No prior knowledge of past game play or points earned per game was considered, and the unskilled participants had to utilize at least 90% of the salary cap provided.
From here, a computer program was created that drafted a single skill based roster for each of the 17 weeks of the NFL season.
Skill-based rosters were set by GLI using a custom algorithm that valued all available players based upon:
- Their weekly salaries, and
- Their fantasy performance from the previous week.
This algorithm was then applied to the game supplied by Star Fantasy Leagues to create a fantasy roster.
In the first week of the season the skill-based roster was selected “randomly” since there were no benchmarks to use for player selection. I put “randomly” in quotes to denote that GLI’s proprietary testing process used certain algorithms to calculate an optimal lineup, but it was based solely on math and not on historical player statistics.
In week two, the same math was used to allocate cap space for position in a lineup, but it also took into account all player performances from the previous week. This process was repeated over the remaining weeks of the season.
The unskilled rosters were set by:
- Compiling a list of all eligible players in the player pool that were not listed as inactive.
- Rosters were then filled utilizing a minimum of 90% of the salary cap space.
For every skilled roster that was created, 1,000 randomly generated unskilled rosters were built, and this scenario represented a single fantasy contest. The skilled roster was then matched head to head against all 1,000 unskilled rosters.
Result: skill dominates chance
After compiling all of the results, the skilled player won just under 70% of the time over the course of the season.
This should go without saying, but I’m going to say it anyway: The difference between skill-based gaming and chance-based gaming is that skill-based gaming is predominated by skill, and chance-based gaming is predominated by chance.
If that’s not clear, let’s put it into numbers.:
- If skill and chance were equal factors in a contest, you would say that a contest is 50% skill, 50% chance.
- If skill prevailed over 50% of the time, it would be fair to say a contest is dominated by skill.
- If chance prevailed 50% of the time, it would be fair to say a contest is dominated by chance.
In these simulations, skill prevailed over chance almost 70% of the time. Add in the human element and we’re looking at an even higher number.
To expand on this point, we took the results of this study a step further to test our theory, and cross-referenced Star Fantasy Leagues player data against the GLI test results.
Over the course of the 17 week NFL season and 35,277 unique valid lineups, we found the SFL customer’s average fantasy football score to be 139.90, which sat in the 93rd percentile of scoring against the 17,000 unskilled rosters.
Further, we took a look at the weekly high scores on Star Fantasy Leagues for each of the 17 weeks of the football season.
Cross referencing this against the 17,000 random lineups that were created throughout the course of the season (and applying our skilled lineup against the 1,000 weekly unskilled lineups), there was only 1 instance where an ‘unskilled lineup’ prevailed against our skilled lineup.
That means 16,999 times out of 17,000, the skilled lineup we provided beat a randomly generated lineup.
In percentages, that means the SFL skilled lineup won 99.994% of the time against the unskilled lineup.
Why did we do this?
The answer is twofold.
Reducing legal ambiguity
One, we did this because the people that built Star Fantasy Leagues love fantasy sports as much as much as the other 41 million – and growing – people that enjoy playing these games in the USA.
We want to make sure that we protect the interests of our customers’ favorite pastime, and protect our interests as a business in a legal environment fraught with ambiguity about how to handle fantasy sports as it relates to gambling law.
Let me help you – this is not gambling against the house. It is a peer-to-peer interaction where a player’s skill relative to the other will, on the aggregate, determine the winner of a contest. And it’s fun.
Balancing innovation and compliance
The second reason was to drive home the point that, in my opinion, not every game touting itself as fantasy sports appears to be a skill-based game, and I’ve seen some that I would be surprised to see fall on the right side of the skill versus chance spectrum if they were put through testing.
If that’s the case, certain games shouldn’t be classified as fantasy sports games, they should be classified as gambling games with a sports theme.
I’m not going to name any names in this piece, and if you ask me in person I’m still not going to answer that question. What I have seen, though, are games that look a whole lot like banking games, games that look a whole lot like prop betting, and games that appear to contravene the most basic tenets of the fantasy sports section in the UIGEA.
What I don’t see is how they resemble the salary cap style fantasy sports games that we’ve crafted.
This is not to say that the concept behind these games is bad or that I’m against their existence; in fact it’s quite the contrary. My point, though, is that when a market becomes more saturated, you’re unlikely to make a dent in it unless you have a unique selling point or a differentiating factor to your business. As the daily fantasy sports gaming vertical grows in popularity – and it will continue to – we’re likely to see new products pushing the envelope of what is legally permissible and calling it innovation.
Innovation and competition are important, but if innovation steps over the line of what is legal and what is not, it puts a black mark on those that work hard to stay firmly on the right side of it.
This is not to say innovation should be stifled, but innovation and timing should be very carefully considered when you’re introducing new products into today’s real money gaming market.
These fantasy sports games are games that get me more excited about watching sports. These are games that keep me glued to a broadcast, even in a blowout, if I’m still rooting on one of my players. These are games that allow me to stay connected with my friends from when I went to school.
These are games that allow me to have friendly competition and banter with people in my workplace. These are games that spawn flourishing Internet communities, connecting me with people from around the world I might not otherwise meet. These are games that allow me to sit on the couch with my father and root with or against him over a beer.
These are games that allow me to fulfill my childhood fantasies of owning and managing a sports team.
Since the skill versus chance debate is such a hot topic, it’s often overlooked that fantasy sports makes for incredible entertainment. Win or lose, I’m in the game when I’m competing. My fan experience is that much better.
As a customer, I don’t care how it’s classified. As a business, the results of this study help to show that it’s classified correctly.
This is one argument that is ready to be put to bed.