The first, and most fantasy-useful project we build with the developer kit.
Using only information about our matchup:
— Drew Brees, Alvin Kamara, Sony Michel, Julio Jones, Keenan Allen, Jared Cook, Matt Prater, LA Rams D
— Russell Wilson, Christian McCaffrey, Saquon Barkley, Corey Davis, Dante Pettis, Greg Olsen, Mason Crosby, Bill D)
Guys we're trying to decide between
— Sony Michel, Phillip Lindsay, Lesean McCoy, Rocye Freeman
And the API, we'll write code that gives us everything we need to make an optimal decision.
Results and Analysis Outputs
We'll look at the distributions of the guys we're considering:
And our team as a whole with each of them:
For close start-sit decisions it can be tough to differentiate between the curves. That's why it's often helpful to look at the raw numbers.
The most important thing we care about is who maximizes our probability of winning, so our analysis needs to feature that heavily.
wp sony-michel 0.505 phillip-lindsay 0.490 lesean-mccoy 0.452 royce-freeman 0.449
The bottom line is Sony Michel gives us 1.5 percentage point increase over our next best option. That's all we really
A 1.5 point increase might not sound like much, but as we'll see in the
league analyzer project
, most fantasy matchups are
close (in 5/6 matchups of the league and week we're looking at, the favorite had less than a 53% to win), so it's helpful to squeeze out every last bit of win probability we can.
Plus this gets us over the all important 50% mark and puts us into favorite status, which is an important physchological hurdle.
Our tool also projects our total team score with each candidate. These are the numeric versions of the plot above.
How often do we start the wrong guy?
A probabalistic approach to Fantasy Football means understanding there's always some risk that things go wrong.
We'll include that in the tool.
First, what's the probability we start the "wrong" guy? For example, if we start Sony, what's the probability Linday, McCoy or Freeman outscore him?
wrong sony-michel 0.573 phillip-lindsay 0.627 lesean-mccoy 0.824 royce-freeman 0.976
We can see its actually more likely to happen than not, but still
likely to happen with Sony than anyone else, which is why we're start him.
This is good to know though, so we don't overreact in the likely event our starter puts up fewer points than one of our backups.
How often do we lose because we started the wrong guy?
Starting a guy who is outscored by someone on our bench is one thing, but how often will this decision cost us our matchup?
For example, if we start Sony, how often can we expect to lose our matchup when we would have won had we started Linday, McCoy or Freeman instead?
regret sony-michel 0.048 phillip-lindsay 0.063 lesean-mccoy 0.101 royce-freeman 0.104
Less than 5% with our top choice, which is good. That doesn't mean it won't be frusterating the 1/20 times it happens though.