Anatomy of an analysis

Recently a new guy showed up on the Handicapper Think Tank forum on sportsbookreview.com. In his first post, he described a MLB system he was using to wager on game money lines. My first reaction was negative as is my general reaction to new systems. But it included a couple of elements that I’ve used in the past and enjoy. So, I contacted him, and he shared some specific information.

The system is in Excel and pulls down the last 30 games for the two teams playing a game. It selects the road results for the visitors and the home results for the home team. It saves these money lines and the results of the past games, 1 for a win and 0 for a loss. He does something that had never occurred to me with the money lines. If you plot money lines on the x-axis of a graph, there is a gap between -100 and +100. He subtracts 100 for each so they are then continuous. For instance, a game with line of -120 and +110 becomes -20 and +10. Personally, I always convert these to the European form. These lines would become 1.833 and 2.100. He then does linear regression on these to develop an equation for each team whereby you can enter today’s money line and it returns a win percentage. Do this for both teams and each has a winning percentage. You can divide each by their total and normalize these percentages to 1.

I decided to replicate this system and back test it. But I had no interest in using Excel because it would be difficult. So, I decided added code to one of my analysis programs. He sent me images of his results for the Cardinal-Cub game on Thursday after the All-Star break. I’m in the process of duplicating the projections for this game. That will be in my next post.