Fantasyomatic vs ELO and CORTANA in the 2018-19 playoffsFebruary 4, 2019 7:47 am
After predicting 64% of all winners in the 2018 season and 56% ATS along with multiple years of success predicting winners in the NFL postseason we wanted to compare the success rate of the fantasyomatic algorithm’s predictions to these other algorithms in the industry and see if FantasyoMatic could again and beat out BOTH ELO and Cortana like they did in the 2017-18 playoffs.
Both system predict every regular season game while fantasyomatic currently only predicts postseason matchups.
We scored each prediction in two ways:
Picking the correct winner
While picking winners is not an easy bet to make money on without a spread, it is certainly a metric for gauging accuracy of predicted outcomes. We do not weight this as heavily as picking against the spread, we did score all three algorithm’s on their ability to pick winners. This was meaningful when it came to picking upsets.
Picking correctly against the spread
We find this to be the most meaningful since this is information that can be used to win money. We score the accuracy of these predictions by looking at each algorithm’s expected outcome and comparing it against the vegas line at kickoff.
So if one algorithm picks TEAM A – 5.5 in a matchup and that team is favored by -6.5 by vegas at close, we consider that a “pick” for their opponent, TEAM B, to “cover”.
In that same example, if an algorithm picked TEAM A -9.5, we would consider that a pick for TEAM A “to cover” 9.5 points or more, etc.
For the 2018 NFL Season, the fantasyomatic algorithm was very competitive in each category.
For your own audit,
ELO on FiveThirtyEight.com: