You know what’s the worst feeling in the world? Scoring like 200 points with an NFL 50/50 lineup that you didn’t put into any tournaments. Like I think I’d rather not leave the house for an entire week over getting first place in a 50/50. That doesn’t really say much because I already stay inside all week, but you get the idea.

My social life aside, I haven’t historically played my cash lineups in GPPs. I do it from time to time if I think it makes sense, but I take radically different approaches to the two league types. In tournaments, I care about both upside and expected player utilization, while cash games are more about median value and gaining access to a high floor of points.

Recently, I’ve noticed that my cash game lineups have been really crushing, even outperforming my GPP lineups in terms of bulk points in most weeks. That’s good and bad; I want a ton of points in heads-up and 50/50 leagues, but I’m not sure if those scores should consistently be my highest.

I’m a huge proponent of testing basically every assumption that we make in daily fantasy sports. There are quite a few beliefs that most players hold as indisputable fact that I just don’t believe to be true, and I’ve formulated my opinions on these subjects by using data to test various hypotheses.

I think the best daily fantasy sports players are the ones who are constantly trying to improve. That evolution as a player is made possibly only if we take a scientific approach to the game; lineups need to be examined after the fact, choices must be scrutinized, and everything we think we know to be true needs to be tested.

With that said, I downloaded my DraftKings game history (available on the ‘My Contests’ page). I sorted my league history into two categories – GPPs and cash games – with the former being anything with more than five entrants and the latter being heads-up games, 50/50s, and three/five-man leagues.

 

My Game History

 

The first thing to note is that there’s a huge disparity between my average cash score (150.7) and my average GPP score (137.0). There are a few different reasons for this, which I will touch on in a bit, but I actually don’t really care too much about my average GPP score. In tournaments, the goal isn’t necessarily to score as many points as possible, but rather to win the tournament. Those two goals don’t always mirror one another.

The real trait that I was looking for in my scores was that my tournament lineups have more upside and my cash game lineups have more access to a high floor. Here’s a look at my distribution of NFL scores over the past few years.

DraftKings GPP vs Cash

You can see how much better my cash lineups have been than my tournaments in terms of scoring points. I’ve actually had a lot of GPP success, though, which I think speaks to the idea that you don’t always need to maximize points in tournaments.

It’s good to see both a higher ceiling and lower floor out of my GPP lineups. My six highest scores and my eight lowest scores ever were in tournaments, which suggests I am indeed taking a high-variance approach to those and a low-variance approach to cash games. That’s a positive.

Another way to view this data – one that is perhaps more telling – is to sort the scores into buckets to see how likely it is that a specific lineup ends up with a particular score.

DraftKings GPP vs Cash 2

 

I want to note that the tail ends of this distribution are skewed a bit because I enter the same lineup into multiple cash games. I had a 200-point 50/50 lineup that was entered a multitude of times, for example, which probably throws off those numbers a bit. The fact that I’ve had a slightly higher chance of scoring between 180 and 199 points in tournaments is probably more valuable intel since that’s a much larger sample of leagues.

Nonetheless, you can see that I’ve been much, much more likely to score at least 160 points with my cash lineups. Over the past few years, 40 percent of my cash lineups have reached 160 points, compared to just 19 percent of my GPP lineups. That’s a massive difference.

 

An Antifragile Approach to Tournaments

 

I mentioned earlier (and I’ve explained often in the past) that, unlike most players, my goal in tournaments isn’t to score as many points as I can. Yes, I know that you need to score a hefty number of points to win any GPP, but I don’t believe the path to tournaments success is through a pure maximization of points.

My GPP strategy is one that emphasizes antifragility, i.e. it gains from disorder. When I’m playing GPPs, I ask myself, “How can I benefit the most if the majority of users are wrong this week, or even if things just don’t go their way?” That means I care a whole lot about expected player usage. I don’t simply select only low-usage players, but I balance expected usage with value to create lineups that I believe will be in the most advantageous possible position if things get chaotic.

The most obvious implementation of this strategy is in baseball, where users often load up on the Rockies (or their opponent) when playing at Coors Field. I almost always fade that game because I know usage will be through the roof. The thing is – playing Colorado at home will lead to the highest scores over the long run because that game is almost always the highest-projected in baseball. The problem is that you’re forcing yourself to compete with so many other users that, even if a team in that game goes off, you’re not necessarily in an elite position.

By using an antifragile approach to tournament roster construction, you’re putting yourself in position to benefit if things don’t go as planned for others. If the Rockies score just a few runs, for example, you’ve eliminated a huge percentage of the field with whom you must compete. That means the winning score is far more likely to be lower than if the Rockies score a ton of runs. Less competition equals higher win rates on more modest scores.

 

A Needed Change

 

Having said that, I still think this data is extreme enough that I can and should consider placing my cash lineups into GPPs. That might seem obvious considering the point differential, but again, I care less about points and more about maximizing my win probability. The tournaments I’ve won in the past have often been with what most would consider “lackluster” GPP scores, which I believe is a good thing because it suggests I’m approaching tournaments in the optimal manner.

Nonetheless, I’ve had enough success in cash games that I’m going to experiment with throwing my cash lineups into tournaments. It will admittedly take a long time to collect the results, too. Since I care only about my odds of winning (or securing a major cash), it will take a decent amount of time to determine if taking a point-maximization approach to GPPs (which I’d be doing by using cash lineups) will actually improve my chances of winning.

The overarching idea, though, is that you can become a significantly better daily fantasy player by testing everything you think and do, even what you consider to be “obvious,” and making changes based on the numbers.