This is a very concise explanation of a startup idea. It will require a background in statistics, discipline, and computer science to execute. Therefore, many details are left out to keep this post readable to folks without these backgrounds. Enjoy.

Recall in my last blog post, we discussed how one can make positive expected value and ultimately generate wealth by studying the game of poker. (I highly recommend reading that before this one) We also mentioned how this was similar to card counting except it is scalable since the house does not limit you as you are playing against others. Additionally, we briefly talked about how this related to sports and financial trading as both have had statistical revolutions where “experts” no longer choose the best stock/player. Following that post, I had a natural question: Where, in the modern day, are heuristics still being used in place of modeling? Secondly, is it possible to model this area? Lastly, for the suckers out there, how can we generate profit?

After many hours of watching YouTube for no other reason than entertainment, I realized there is a viable market. Fantasy Sports. Fantasy sports are dominated by so-called experts such as Matthew Berry, that entertain folks on ESPN by giving their heuristic-based opinion. Sure, like baseball, there are some basic statistics to “measure” or “predict” fantasy points. But “experts” never cite any statistic related to your team's expected wins.

So, fantasy sports are dominated by heuristic-telling experts that lack a mathematical foundation. Next, is it possible to model this area? Well of course it is possible. As we previously discussed, my friends and I have generated lots of expected value (EV) by betting profitably against sportsbooks. Therefore, it is already possible to create models for statistics such as yards, goals, touchdowns, home runs, etc. Naturally, all one has to do is transform this into expected fantasy points.

Lastly, how can you generate EV? Well, there are hundreds of different contests and this is the core idea of the startup. One needs to build models to accurately guess the expected value of each player relative to the contest rules based on their draft position. Then, if one can consistently take players above their draft position, they will be able to quantify their expected wins above average and therefore their expected profit margin.

Where can things go wrong? Ok, there are a few fundamental challenges that make this idea an interesting one for someone to explore. First, there are so many different types of fantasy contests and therefore choosing the ones that are most profitable will certainly be challenging. However, I suspect it will be much easier to model daily or weekly contests rather than season-long ones. To explain this, I would like to refer to predicting the weather. Weather, as models define it, is a chaotic system. This means that slight perturbation in the weather today will drastically affect the weather down the road (see the Butterfly Effect by Lorenz). Therefore, it will be very challenging to predict the weather a year from now, but the weather tomorrow can be reasonably anticipated. My prediction here is that fantasy sports, like weather, is a chaotic system and therefore very unpredictable over the long term. (This is backed by the idea that future lines such as MVP or DPOY are more heavily juiced than money lines for tomorrow’s game). Thus, I predict one needs to identify short-term contests that can be accurately scaled for maximum edge.

For now, this is all I will say about this idea. I predict that fantasy sports will undergo a revolution similar to poker in the near future and so this idea will probably become less profitable over time. However, feel free to try and execute it yourself. Personally, I probably will not touch it because like the title says, all this idea really does is print money. There is no impact on the rest of the world. But money is important to raise a family and enjoy subtle things in life such as travel, so I understand folks who are interested in printing it.