Consistency and Cash Games

One of the most overrated stats in all of football is yards per carry (YPC). The stat is pretty much useless because it’s so affected by outliers. A running back can be having a poor game of 15 carries for 45 yards (3.0 YPC), then break off a 70-yard run that will catapult his average to 7.19 YPC. All of a sudden, he “ran all over the defense”—a conclusion that might result from one broken tackle or a defender falling down.

Is that 115 yards on 16 carries really the same as a running back who continually gashes the defense for seven yards? Of course not, and it’s vital to understand that difference when selecting your fantasy lineups.

In his book Antifragile, Nassim Nicholas Taleb makes a distinction between the resilient and the antifragile. The resilient can withstand shock; it remains the same in the face of outliers. The antifragile, on the other hand, not only withstands variance, but it prospers from it. The antifragile improves with chaos. In many ways, your daily fantasy lineup selection should be built upon a similar foundation. Outlying poor performances will hurt you regardless of the league type, obviously, but you want your head-to-head lineups to be as resilient as possible; you want consistent play from every position. When a lineup in the 60th percentile is a winner in 50/50s and most head-to-heads, you want little variance.

Understanding Long-Term Trends

In any daily fantasy sport, you have a lot of decisions on your hands, the most overlooked of which might be salary cap allocation. The individual matchups are of course important and every lineup should be built upon specific information relative to that day’s games, but I think some players overlook the importance of long-term trends. While the value of an individual player is fleeting, the importance and consistency of specific positions is more everlasting. Are quarterbacks generally safer options than running backs? Is it ever okay to pay up for a kicker? Which types of wide receivers provide consistent production?

NFL Head-to-Head and 50/50 Strategy

In both head-to-head and 50/50 games, you want consistency. If you average 150 points with half of your lineups scoring 100 points and the other half scoring 200 points, you’re actually not going to be a profitable heads-up player. If you could find a way to score around 150 points each time, however, you’ll be nearly unbeatable over the long-run.

To back up that idea, let’s take a look at some more DraftKings data on the average scores in different league types.

This is awesome stuff. You can see the average top tournament score (197) and average score that finishes in the money (157) dwarf the numbers in head-to-heads and 50/50s.

Looking at those head-to-head and 50/50 leagues, the average top score in the latter is much higher than that in the former, which is to be expected since there are just more lineups (sometimes many more so) in 50/50s. Nothing strange there.

But here’s what’s most interesting to me. The average “in the money” score in head-to-heads (143) is three points lower than that in 50/50s (146). Since the top half of entrants get paid in both league types and we’re dealing with a huge sample size, you’d expect the numbers to be equal. You’ll have more outliers in a 50/50 since there are more lineups, but if you took the same sample of heads-up lineups, you’d think that the score distribution and average would be the same.

But it’s not. Further, despite a higher average “in the money” score in 50/50s, the average overall score is one point lower than in head-to-head leagues. That means the deviation between the average score and the average winning score in 50/50s (17 points) is a lot higher than the same deviation in heads-up matches (only 13 points).

Here’s my explanation for the difference that, if true, could really alter the way you enter both league types: people are approaching 50/50 leagues with the wrong strategy. It initially seems like 50/50s might be more difficult since the average “in the money” score is three points higher than in heads-up matches, but I don’t think that’s the case.

Instead, I think many daily fantasy players are approaching 50/50s with a high-variance strategy much like what you might seek in a GPP. So there’s a wide gap between the best scores and the average scores that increases the overall average, but the outlying top lineups might be throwing off the mean.

Here’s a visualization. First, this is how scores might be distributed in a 10-man 50/50 league in which players submit optimal lineups.

This sample distribution matches the DraftKings data—an average “in the money” score of 146 and an average overall score of 129. The dotted line represents that average. You can see that half of the 10 lineups finish above that 129 mean, which is what we’d expect if players submit lineups as they should with a low-variance strategy in mind.

However, this is an example of what we actually see with 50/50 lineups.

Again, the average of the “in the money” scores is still 146 and the overall mean is still 129. But there’s a larger deviation in scores, so the top lineups make it seem as though players are better when they’re really just submitting high-risk/high-reward lineups.

Now, here’s the important part. Take a look at the average lineup dotted line. It’s still at 129, but now there are only four lineups that fall above it. The fifth-best lineup—one that would cash in this 50/50 league—has just 119 points, which is 10 below the overall average.

So despite the higher average score in 50/50s over head-to-heads, the deviation in points suggests you might be able to cash in them more easily. Daily fantasy players might see the large number of entrants in a 50/50 (which can be huge at times) and automatically think they need a high-variance lineup with lots of upside. That creates a phenomenon through which below-average lineups can sneak into the money. In reality, you should approach a 50/50 just like a head-to-head league, trying to find consistent, sustainable production.