Thursday, Sep 23rd

Understanding and Finding EV


Expected Value (EV) is something that is constantly brought up in the fantasy and DFS worlds. It is a concept that is very relevant to DFS, but also one that is very misunderstood by many — including a lot of the people who use it to give advice. The first thing to understand is that “EV” means almost nothing if your sample size is not large enough, or if you are regularly making significant changes to your approach and playing style

Here is a brief example of how EV is actually calculated. Let’s say you want to buy XYZ stock for $100, and we know what the probabilities are for various levels of return on it over the next year. Obviously, there would be many more possible outcomes for a given event, but for simplicity we will just break it down to five possible outcomes: 

  • 20% chance the company folds and is worth $0 in a year
  • 30% chance of their stock being worth $50 in a year
  • 20% chance of their stock being worth $90 in a year
  • 25% chance of their stock being worth $200 in a year
  • 5% chance of their stock being worth $1,000 in a year

To calculate the EV, you would do the math on each of those outcomes and add them together:

  • 20% x $0 = $0
  • 30% x $50 = $15
  • 20% x $90 = $18
  • 25% x $200 = $50
  • 5% x $1,000 = $50

In this example, the Expected Value of XYZ stock would be $133 ($0 + $15 + $18 + $50 + $50). Since the initial price of the stock is $100, your profit over that year would be $33 (which is a 33% return). Since that is a very positive outcome, it means that buying XYZ stock would be viewed as having Positive Expected Value, or “+EV.” Obviously this means you should BUY, BUY, BUY!!! It is “+EV”; what more is there to think about?!? We did the math and everything; now all you have to do is click the button and the money starts flowing in, right? Well, not so fast…..

The first thing to consider is that in this example, there is a 70% chance that your investment will lose money. You have more than a 2-to-1 chance of this decision costing you money. This is something that is hard for our human brains to reconcile — this decision will probably be a bad one, but it is the right one. For our purposes, this is why you will often hear people say “if you are playing GPPs, you are going to lose a lot.” This can be a dangerous line of thinking, however, because bad decisions and strategies can easily be written off and attributed to variance that will “eventually even out”…but that is no guarantee.

The overlooked thing about EV, and why the way many fantasy “experts” talk about EV is misleading and sometimes irresponsible, is that it is a projection and estimate. The inputs are guesses, which means the calculations are guesses as well. The numbers that are input are difficult to quantify (and in the case of DFS, almost impossible). To show you what I mean, let’s go back to the example of XYZ stock, but slightly tweak the numbers:

  • 30% x $0 = $0
  • 20% x $50 = $10
  • 20% x $90 = $18
  • 29% x $200 = $58
  • 1% x $1,000 = $10

Using these projections, the EV of XYZ stock is $96 (0+10+18+58+10). This means it is Negative Expected Value “-EV” and your expectation would be to lose 4% on your investment. That’s almost a 40% difference in your EV (+33% to -4%) just from some very slight changes in probabilities. And the reason this is critical to understand is because THE PROBABILITIES ARE GUESSES!!! We don’t know for sure what the odds are of any of these events happening, they are just somewhat educated guesses on the likelihood of what COULD happen.  

For DFS purposes, these projections are even more fragile. We are guessing on the probability of players playing well and also guessing on the probability of how decisions, strategies, and approaches will pay off over time. Any math equation consists of both variables and constants. Variables change and can be unpredictable while constants remain, well, constant. Obviously there are things we can’t control or know in DFS, so the key to maximizing our EV is to have systems and processes in place that will give us as many constants as possible. If you are changing the contests or amount of money you play each week, changing how you select players each week, doing lots of research some weeks and barely looking other weeks, or any other things that are drastically different on a week to week basis, you might as well be playing slots or scratch offs because you are introducing so many variables to your DFS play that it is nearly impossible to expect to realize any sort of “EV” over time.  

I am a big fan of the Netflix show “Ozark.” In the 3rd season of the show, the main character, Marty, has a great line:

“One person making one bet, I couldn’t possibly tell you what they’re gonna do. But the law of large numbers tells me that a million people making a million bets, that is completely predictable, completely ordered.”

The lesson here for us is to keep making our smart, calculated bets — but to have a process and expectation of ALL ASPECTS of how we are making those bets and decisions. People will say or think things like “It’s +EV to play RBs who are home favorites” but basic things like that are not nearly as important in realizing EV as structuring lineups, use of bankroll, contest selection, etc. This course will dive into how you can approach DFS in optimal ways that will actually improve your expected returns and are likely to be realized over short and long term time horizons.