Hey Everyone,
last year I had a thread running for anothr system that worked, but didnt work as well as it did the three years before. During the year I read into Sabermetrics more and more and came up with the idea to try and build a database with all the needed statistics to try to pick winners in this years MLB season. It took me quite a while to get everything going (excel-file is about 60 MB) but I am finally done as of now and will get my toes wet during spring training.
I highly encourage NOT to follow at this point in time, I use covers as a place to track my picks and find out how good (or bad) my sabermetrican approach will work. I was inspired by the work of Sports_Network but made some minor changes to his approach, I might get into more details later on. The MAJOR difference to his thread will be that I will post the attached units before the games start :-)
Anyway, the general approach is very similar:
1) I will use sabermetrics to determine the (based on historical data) possible outcome of the game with two approaches: a plain sabermetric approach with BsR and RC. The second approach is a monte carlo simulation, based on historical data AND is including all the different possible outcomes of EVERY at bat, basically simulating every game about 100.000 times. The monte carlo simulation should be ready to go by the end of march. The scores from these two approaches will then be translated into winning% and then into odds.
2) Based on the outcome of the approaches above, I will then have a couple of different posibilities to bet.
a) Big favorites will be run line plays, units depend on the difference in a) runs as of my model and b) the odds from the bookmakers.
b) There will be quite alot of games where the wrong team will be favored due to public attention etc. These games will be straight ML picks and will only be played on dogs obviously.
c) There will also be games where the favorite is correct, but due to public attention the price of it (the line) will be much too high. These will be pure value plays on the dogs, as most of the time the favorite will win but at unattractive prices.
4) Over/under bets will also be included, although this is the area I have spend the least time on as of now. This is where I will work on now to be ready when Spring Training kicks in for real (I hope I will have this adressed by the end of the weekend).
As you can see there will be quite some battlefields and I also have to state that there will be more often than not a schedule full of games for me, as my backtesting showed me that (leaving O/U aside) during a normal day with 14 games I will probably have 6-9 picks.
Once again, I do NOT recommend to follow right from the start, backtesting was nice but picking games BEFORE they start is a whole different animal. I cannot say at which day my picks will start, especially in Spring Training its hard to get the lineuos of the games PLUS the pitchers will change that often that its hard to run a simulation on that.
Anyway, my goal is to make 100 UNITS per season, my plays will be for 1,2,3 and 4 UNITS and this is all you need to know as of now :-)
I wish all of us a sucessful season and am happy to discuss the games as well as the approach itself.
Sabermetrics!! 
Hot Forum Topics
