There’s a weird trend in statistics that we see every night in DFS – something random happens, like the 10th man on a team getting an extra two minutes because of an injury or foul trouble, getting hot, and riding that to a big, weird fantasy night. So, what are the odds of that happening? That exact situation? Pretty small. But we also see situations like this happen seemingly every night – it’s not the same 10th man, but almost every slate has some weird, statistically improbable event happen. What gives?
Wyatt Earp lived in America in the late 1800s during the American Old West era and is now famous contemporarily because of an interesting character trait: his ability to survive a large number of gun duels. He reportedly won hundreds of them, and while they do rely on skill – notably, how fast and accurately you can draw your gun – the odds of him coming out on the right end of that many duels is super low. Even further, it’s reported that Earp rarely got the first shot of the duel, as he relied on accuracy over speed. That’s fine and all, but that makes his duel win rate even crazier – you’d think that if over 100 guys got a first shot at him, even just one bullet would accidentally hit him.
In statistics, weird things happen all the time. Are they probable? No, but in large sample sets, they don’t have to be. Even if something happens two percent of the time (most would consider that low), that something happens 20 times in a 1,000. It’s small, but it’s not zero. And if you think about it, statistically improbable things by definition have to happen because they’re statistically improbable, not statistically impossible.
Take a coin flipping contest. You get 10,000,000 people in a room and you match them up in a tournament – a player “wins” if they flip heads and “lose” if they flip tails. After 20 rounds, there has to be contestants that are still in the competition – by definition of the competition – that have flipped 20 heads in a row. The chances of that happening if you just sat down in your room and flipped a coin is roughly 1 in 2,097,152 (well it’s a bit more complicated than that according to this, but that’s a little more intense than we want to get into in this DFS article). It’s statistically improbable, yet it’s statistically impossible for it not to occur because of the nature of the competition.
I like this post I came across in a math forum (don’t judge):
Here is an important point about probability. If there are enough possible events with very low probability then it is likely that at least some of them will occur. Or, if an event with low probability is given a large number of opportunities, then its chance of happening at some time may be quite large. What would be really unusual is if nothing unusual ever happened.
Consider this example. Let’s say that something has a 1.0E-10 probability. If there are 1.0E+10 such independent events, then the probability that none of them happen is only about 1/e, or about 0.37.
If you judge each situation independently, you’re right to think that something is statistically improbable and thus you shouldn’t incorporate it into your DFS research. However, it seems like this is an important part of DFS that we miss – random events will occur. As said above, what would be really unusual is if nothing unusual ever happened. The question then for DFS isn’t whether random events occur; rather, it’s how to prepare for these random events and profit from them.
Nassim Nicholas Taleb has popularized a term called ‘’antifragile” that Jonathan Bales has adapted for the DFS world. The point is that, especially in GPPs, setting ourselves up to benefit from chaos is an incredibly successful, albeit stressful at times, strategy in DFS and investing in general. This point applies to this article – if we know chaos exists (I’m calling statistically improbable events chaos here for the purpose of merging these two ideas), we should put ourselves in a position to benefit from it. It really comes down to whether you want to win contests or maximize points – if it’s the latter, you should probably always try to take the players with the highest-projected point total. However, if it’s the former, then in GPPs you should keep in mind that going chalk often doesn’t get you anywhere. It can provide a nice floor, but you’ll have to hit on low-owned players that go off in a seemingly improbable way.
The issue, of course, is how to identify and find these chaotic, statistically improbable events. Aren’t they by nature almost impossible to identify ahead of time? The answer is yes, but the nature of DFS, GPPs, and managing bankroll is that you don’t have to perfectly identify them often. In fact, it can be a rare occurrence – even as rare as the event itself. However, when you do predict them correctly, you’ll be at such an advantage over the field that it will easily make up for all the times you were wrong. If there was only one possible random event and you had to predict it just once, you would be foolish to attempt that feat. However, since you already know that something improbable will occur, you can take calculated big risks in GPPs. And every once in a while, they’ll pay off in a big way.
