It’s no secret that the internet is full of fantasy advice.  You can easily find advice in articles, in podcasts, on TV/video, and on Twitter.  If you know where to look, you can find advice that can be extremely helpful, but there’s one giant piece of advice that you need to take away from the mountains of opinions and analysis at your fingertips: take it with a grain of salt.
What do I mean by this?  At the end of the day, a fantasy analyst cannot give you certainty, but only probabilities.  No fantasy expert” can predict the future with total accuracy, no matter how emphatic they are about their picks.  There are simply too many variables and unknowns which influence individual player performances on any given day and throughout the season.

Don’t believe me?  Let’s go back to the beginning of the 2014-2015 NFL season.  Within the top three running back picks that analysts universally touted was Adrian Peterson, who lasted a total of one game before involuntarily hanging up his jersey for the rest of the season for a now-infamous off-the-field incident.  Before the season began, no expert in his right mind would have predicted that a far weaker player like DeAngelo Williams would outproduce Peterson in fantasy points for the year, but that’s exactly what happened.

Like injuries, suspensions are things that no one can really predict, so no one can be blamed for not predicting them.  After all, they happen out-of-the-blue and therefore cannot be predicted, but that is really the point here: experts can give advice based on probability, but not on certainty.  There was a high probability that Peterson would outperform Williams before the season began, but there was never a certainty that this would happen.

But what about things that aren’t mere happenstance, such as player regression or breakouts?  Aren’t these things that fantasy experts should be able to predict?  Well, yes and no.  It is reasonable to predict some things with certainty, such as predicting that players will improver after their rookie seasons, or decline as they approach typical retirement age for their respective sport.  But as we know, some players buck these trends.  Opportunities change.  Personalities mesh or clash which affects playing time.  Coaching schemes are modified which helps or hurts players.  There are many variable that can cause specifics to fall out of line with generalities.

At the end of the day, we can’t have total confidence that fantasy analysts will be correct in their opinions.

Does this mean that fantasy advice should be completely dismissed?

No, not at all.  Advice from fantasy analysts should be understood as helping not to determine certainties, but probabilities.  In other words, it can help you figure out whether Player A is more likely than not to outperform your next-best option in Player B.  Probabilities work better when the sample size is large enough, so season-long rankings are more likely to benefit you than week-to-week of daily rankings.

Using probabilities in your player selection

Imagine this scenario: you are about to enter the draft in any sport, and you have the following data on hand which tells the likely floor and ceiling of each player:

% likelihood of scoring/X number of points

Player A – 75%/50 pts, 25%/75 pts
Player B – 65%/40 pts, 45%/70 pts
Player C – 75%/20 pts, 25%/90 pts
Player D – 85%/15 pts, 15%/125 pts
Player E – 90%/10 pts, 10%/130 pts

So basically, Player A has a 75% chance of reaching at least 50 points, and a 25% chance of getting 75 points or more.

If you had to build a team starting with two of these five players, which would you pick?  I can tell you that if I wanted to build the team with the best chance of the highest score I could possibly get, I would choose A and B.  Why?  Because I’m betting on players who are most likely to score higher than the others.  Will they necessarily score the most?  No, because as we already established, there is no certainty in player predictions.  We could hypothetically choose two players at random from this group that outscore all other player combinations.  As fantasy drafters, our task is to increase our likelihood of scoring higher rather than pick the players who will score higher, since this is impossible to know.

Why does any of this matter?

If you know that you are letting probabilities guide your way, you should know that:

• sample size is important
• historical performance is a better guide than forward predictions
• you don’t have to hang on every word of fantasy analysts
• you may have the optimal team in your league but you will still lose in at least a few weeks of the season