We are now one month in to the MLB Regular Season and it is time to take look what have happened so far on the field and in the betting account.
This April have not been different to previous years in that we have seen some big surprises. In writing moment Astros are leading AL West with a record of 15-7 and currently winning their last nine games. Not a huge surprise, but maybe a bit interesting, is that Royals have been able to continue from where they left last season and are playing better than (at least I) expected. New York Mets have also started slightly better than expected and are sitting atop of NL East with 15 wins and 7 losses.
From a betting point of view the month ended with mixed feelings. Moneyline (ML) picks continued to hit with a pretty good rate and where about +15 units for Regular Season games (+22 units in March and April Spring Training games). Runline (RL) picks where about -5 units and Totals picks -24 units. The complete past performance overview can be found here.
Going forward I feel very comfortable of how The Handicapper produces both ML and RL predictions. This season there have been 284 side picks (ML and RL) so far and even if this is not a big enough sample size to draw any kind of conclusions of longterm profitability, it is a big enough, to make me feel comfortable. With Totals picks the situation is not that bright. The sample is only 117 picks so it could just be bad luck, normal variation or some other kind of noise. My gut feeling though, that it might be a question of something else, and the situation won't correct itself without some kind intervention. The full season simulations I ran before the season predicted that there would be about 8.15 runs scored per game this year. So far we have seen 8.52 runs per game, which is a significantly higher number than predicted. In April 2014 we had 8.42 and in April 2013 8.46 runs scored per game so it might also just also be a 'first-month-of-season' or 'April' effect. How the model (The Handicapper) uses park factors might also have had an impact on the bad result. Park factors are volatile on a year-to-year basis and therefore the model uses five year averages. For example Houston Astros Minute Maid Park has a league average (100) run scoring park factor. So far this season (minimal sample size) there has been on average 5.5 runs scored in games played at the ballpark. If a run scoring park factor would be created based on only current season stats the Minute Maid Park would have a number somewhere between 50 and 60.
Until I figure out if there is a problem with how the model uses park factors or if they have to be used with a somewhat different approach, I am going to be very selective or lower the bet size with Totals picks.