The man behind the model
My name is Philip Klafta and I am the creator of OutlierProjections. I built this model during the pandemic with the goal of creating the most accurate information in the industry. I recently graduated from the University of Chicago with a graduate degree in Public Policy and am currently working in consulting. My favorite card is the 1990 Topps DionSanders
.png)
Results Based Predictions
I am nothing without my results. The graph on the right shows results from the 2020-2023 MLB season (as of mid May '23) in which my projections predicted >5% difference from the consensus price at sportsbooks. Of the 7,200+ games over the last three and a half years the model identified an edge on 1,630 games and went 733-833-14 for an ROI of 4.34%. This graph will be updated at the end of the year or can be tracked daily at https://myaction.app/The_Outlier.
Why is this model successful?
The win probability model, which is the flagship of the page, is based off of predictive stats like xwOBA, wRC, and xERA These stats are developed using mainly Statcast launch angles, exit velocity, and controlling for park factors which create a more true outcome for each event. For example, when a weak ground ball is hit using traditional data the out come is binary (either 1 for reaching safely or 0 for an out). However, what xwOBA, wRC and xERA do is turn each event into a probability of resulting in reaching safely. So, for that weak ground ball, it is significantly more valuable for us to calculate that as .19 of a hit rather than the binary 0 or 1.
Disclaimer
It is VERY important to recognize that this model does not take into account weather. It does not calculate for rain, wind, or temperature of the game.
.jpeg)
