MLB Monte Carlo Simulations

[mickeyg13]

Hello from a n00b here. Although I've played my share of poker, I never got into sports betting (I don't like gambling if I'm -EV). Recently though I've thought about developing a system using Monte Carlo simulations to estimate each team's "true" odds in each MLB game. I'd then compare those numbers with the odds given by the money line and bet accordingly. However, before going through the effort of creating such a system, I'm curious as to whether or not others have had success with similar systems. In other words, I suppose I'm interested in how much the money line tends to deviate from the "true odds" while I'm also interested in how accurate such simulations can be at estimating the "true odds." I'd appreciate some legitimate responses in addition to the inevitable flaming of the n00b.

[dudeimstoked]

thats what i do in my system. it takes a ton of work however. good luck.

oh, and MLB is pretty much over so you might want to devote your time to basketball or do this next year.

[mickeyg13]

I know it would take a ton of work. My intent was never to use it for the current season. I need a lot of time to develop the system and sufficiently test it. The reason I would do it for MLB is because the numbers give a better representation of what actually happens (plus the fact that I just plain enjoy following it the most). My gut tells me there is potential for a lot of exploitation. That said, I would rather not spend the whole winter programming only to find that my system was doomed from the start. If you don't mind, do you have a ballpark ROI?

[ImStillBen]

What an original concept.

[Post-Oak]

[ QUOTE ]
What an original concept.

[/ QUOTE ]

Please ban this troll.

[SunOfBeach]

With all the new saber-modelers out there, these types of things are becoming less and less valuable. While a good mathematical model can still give you an edge in baseball, it's just no way near the edge you could have had 5-6 years ago doing the same thing.

Read Tom Tango's "The Book" before you begin anything like this.

[mickeyg13]

[ QUOTE ]
What an original concept.

[/ QUOTE ]
As I am new to the forum, I actually thought this was an original concept. I haven't had a chance to read through all the archives, but a quick glance seemed to show that some have tried this, so I was wondering it it worked out for them.

Would it be fair to say that the best simulation systems struggle to have say a 3% edge?

[Thremp]

[ QUOTE ]
Would it be fair to say that the best simulation systems struggle to have say a 3% edge?

[/ QUOTE ]

This is a really vague claim. If you're talking about every game on avg v WA lines, then yeah almost impossible. If on the other hand you're talking about being able to sustain a ROI of 3% against actual lines using a Kelly wagering system. Then yes obviously.

[TomG]

Your biggest challenge will be to determine which statistics best predict a player's future performance. For example, how do you best measure a player's offensive power rating? How do you measure his defensive contribution? What about pitching? What about the bullpen? How should you account for home field advantage? How will you stay current on a team's line-up and injuries? I recommend you spend a lot of time reading articles on Baseball Prospectus, Hardball Times, and Sabermetric sites. It's a lot of work, but if you enjoy the challenge you'll learn a lot and find it very rewarding. When I was doing my "Robot System" I derived a lot of enjoyment out of calculating a money line that happened to fall within a few cents of the book's posted money line. Good luck, don't let others discourage you.