The only way to bet money line sports, imo, is to find the true odds and compare to the posted odds with the casino or website you are playing with. This method is not one of predicting the future but merely finding an imbalance in the market. If we can exploit that imbalance over time we should be profitable. But that is all dependent upon the accuracy of your true odds.
For example we know that flipping a coin is a 50/50 proposition. If someone was willing to pay you $5 for every head that comes up and you pay them $4 for every tail you wouldn’t need to take out a calculator to figure out you were getting the best of this bet. But if you did you would compute your expected value to be 5*50% – 4*50% = +.50. So you would be expected to win $.50 every time you flipped the coin. Short term of course you could lose but over time statistics would win out and you would be profitable. This is exactly how casino table games work. Generally they have a 1-5% edge on table games and over time they are guaranteed winners.
Baseball might be the ultimate ML sports bet in the US. Using a trend based model like in the NFL is not as easy in MLB because in baseball it is all about the guy on the mound. Sportsbooks and bettors are not as easily swayed by a good, or bad, performance in baseball as they are in football. A good example is 5 minutes after the Super Bowl, a game in which nobody seemed to like Seattle, all of the talk was whether Seattle would be a Dynasty. The players hadn’t showered and they went from an underdog to one of the best teams of all time. The truth generally lies somewhere in between. That over reaction in the NFl is exploitable. In baseball we need to take a different approach.
For the last two seasons I have been looking to develop a model that gives better odds of a teams success in a particular game. The model looks at the lineups and pitching matchups closely. Any change in the lineup can cause the model to move the odds significantly, depending on the player. For instance the model has the following odds on todays Detroit/Dodgers game
Miguel Cabrera is in the starting lineup. But lets say he is benched in favor of Don Kelly. How much would you move the line? Well the model would put Detroit as a +112 underdog. A 12 cent move. That may or may not seem like a lot to you but it is fairly significant. Often times lines will not move as much as they need to when a lineup is different than the teams standard lineup. The model tries to give the best odds for that lineup.
When betting any sport it is important to get the best line possible. None more so than baseball where you are taking a ML. Here is the math.
Lets say you play 300 games this season at an average of -150 per play. You are a $100 bettor and you win 60% of your bets. At the end of the season you would be flat. You would have won 180 games for +18000 and lost 120 games for -18000. Your buddy plays the same games as you and he has a book that has much wider spreads. He averages -155 on the same games. He would be +18000 and -18600. He would have lost 6 units over the season because he was not playing at a favorable price.
I will be posting my models true odds, when I can, once the lineups are out. If you need a specific game tweet me @statsational and I will try and get you a number with a presumed lineup. @MWPGT will also be posting his plays periodically so be on the lookout for those.