Using very basic stats for both leagues and building models that can predict the game outcomes. The models are built using XGBoost, sckikit-learn's GaussianNB, and NLP.
At the end of the day, since the inputs are pretty basic stats, the model accuracy between the various implementations does not very as much.
NBA regular season data obtained from 2014 through 2019 season. NFL regular season data obtained from 2014 through 2019 season.
NFL 2019 ~ 68% NBA 2019 ~ 66%
-
Need to do more research what other features to use to improve
-
Any way to take into account team make up (# of all stars?)
-
current strategy
- choose games that are above certain threshold of confidence
- bet on those, compare outcomes, accumulate money
- describe strategy by name
- return strategry wins/losses?