Here’s Why I Think People Get the Upside Calculation All Wrong
I’ve had sort of a nagging feeling for a few months that I disagree with almost everyone when it comes to how to think about upside in the context of fantasy football. For instance, in the mid to later rounds I’m often taking guys that everyone has forecast for steady production, while other drafters are often taking “high upside” guys. The discussion on this choice always boils down to “yeah, but upside”. For example, Friday night on Twitter I said it was illogical for Aaron Dobson to have a higher ADP than Greg Little. I got a few responses that were essentially “yeah, but upside.” I don’t think that’s correct.
Let me actually back up and talk for a minute about what value is in fantasy football. A player’s value is probably approximated by the formula:
PV = (%Likelihood Player is Top 24 at Position * Value of a Top 24 Player) + (%Likelihood that Player is Top 12 at Position * Value of a Top 12 Player)
Which is to say that you know that some players that could be every week starters for you have little chance of actually cracking the top 12 at their position (maybe Lance Moore?) and yet their low percent chance likelihood at getting into the top 12 doesn’t mean that they have no value. Those players still have some value.
In general I think people often discount the value of a top 24 player, although that’s not my quibble in this post. My quibble in this post is that I think people get the calculation on both of the input variables in the above equation wrong.
Let’s start with the %Likelihood Player is Top 24, just for simplicity’s sake. You have to be able to figure the odds on that as well (even if just intuitively) in order to make good decisions. Most people would look at a player like Greg Little and assign 0% chance that he gets into the top 24. Then they would look at Aaron Dobson and assign some chance that he gets into the top 24, which would lead them to take Dobson over Little. Except here’s the problem with that. The Sim Score app tells us that 6 of Little’s 20 comps went on to have the equivalent of a top 24 season (essentially over 850 yards). So Little might have as much as a 30% chance of having a top 24 season. I wouldn’t say exactly 30% chance because I think that’s being overly specific. But he has some chance.
So then what of Dobson having a top 24 season? First, we know he’s not starting right now. That impacts the odds. The conditional probabilities are affected by Dobson’s place on the depth chart. Dobson’s chances of getting into the top 24 if he isn’t starting are probably so low as to not be worth talking about. Then we have to calculate the odds of Dobson winning a starting job and multiply that number by his odds of finishing in the top 24 if he did win a starting job.
DOBSON TOP 24 ODDS = ODDS HE STARTS * ODDS HE BECOMES TOP 24 IF HE STARTS
So what are Aaron Dobson’s odds of becoming a starter in NE? I think 25% is probably a decent guess. The odds are that Kenbrell Thompkins keeps the job unless he becomes injured. Then even if Thompkins does get injured, Dobson would have to compete with Josh Boyce. That means we would have to multiply that 25% by whatever Dobson’s odds of finishing in the top 24 if he did start. Conditional on being a starter, does Dobson have a 50% chance of making it into the top 24? That seems about right. He would likely still be behind Amendola (and potentially Gronk/Sudfeld) for targets. He would also be a rookie, although he would be in a high powered offense and that would help.
If you do the math using the numbers I suggest above, you get about 12-13% chance that Dobson finishes as a top 24 wide receiver. That’s compared to maybe 30% or so for Greg Little. Those numbers are just “back of the napkin” type analysis and you might disagree with them. But if you actually go through the steps I think you’ll come out close to where I am on this. I think the key thing is that most owners aren’t going through the steps. They regularly do things look at Marques Colston and say 0% chance to finish top 12 without consulting information that would inform the probabilities. They look at Michael Crabtree last year (one of Greg Little’s comps by the way) and say 0% chance of finishing top 24, so they take someone else instead.
I think the most exploitable inefficiency in fantasy football right now is that a lot of owners mistake unlikely for impossible and they similarly mistake likely for certain. That’s what they’re doing with Greg Little right now. I don’t think it’s likely that Greg Little finishes the year as a top 24 WR, but fantasy drafters are assigning it a close to zero percent chance and I think that’s wrong. All of their miscalculation is expected value (EV) that they’re leaving on the draft board for you to clean up.