MLB Betting: Using the Log5 Method to Predict MLB Statistics

SBR Staff

Saturday, May 30, 2015 12:25 AM GMT

Calculations like the Log5 formula can be the starting point or core of any baseball betting system or simulation model.  It can be extended further with the use of additional advanced research.

Using the Log5 Method to Predict MLB Statistics
I hear a lot of baseball handicappers and even MLB managers cite specific batter vs. pitcher matchup statistics to help them make decisions.  The truth is that most of these decisions are based upon sample results that are not predictive.  For example, if a batter has 7 hits in 21 AB’s in his lifetime against a certain pitcher, that small sample should not be used to predict success or failure.  The technical term is that this sample size of AB’s is not statistically significant.  Since so much luck is involved in an individual in plate appearance, placing any weight or consideration in a small sample size would not be beneficial in handicapping and/or managing a baseball game. Because most hitters will typically have a small sample history against certain pitchers, alternative options for predicting hitting success or failure have been created.

Bill James, one of the “godfathers” of Sabermetrics and advanced baseball statistics, devised the Log5 formula to address this situation.  The original intention of Log5 was to determine expected win probability between two teams with respective winning percentages.  Log5 can actually be extended to statistics beyond expected win probability.   If a hitter with a .290 average is facing a pitcher with a batting average against if .270, what should be the resulting expected batting average?  The Log5 formula is used to predict the probability for a hitter with a specific batting average (BA) of getting a hit against a pitcher with a specific batting average against (PA).  In order to project this with accuracy, the league batting average (LA) will need to be referenced.  In short, the formula will be drawn like this:


Assuming a league average of .250 and if you plug in the statistics that were referenced above, the expected batting average would be .312. 

As was previously stated, the Log5 formula is dynamic and can be extended to other key baseball statistics.  For forming my MLB picks, I will use this formula to project walks, home runs, and strikeouts.  Additionally, projecting on base percentage (OBP) is the core of many simulation models.  In short, the Log5 applied to OBP will project the probability of a hitter’s success of reaching base. 

How can this be specifically applied to your handicapping? Whether you make your own stat projections or use current stats to assist with your handicapping, these numbers can plug into Log5 to help project winners. 

Let’s assume that we make team projections for OBP.  For example, let’s assume the following:

If you just take a look at the OBP stats, you would notice that the Away team has a superior OBP probability.  Without factoring in anything else, you can assume the Away team has a higher probability of scoring runs.  At the same time, the Home team’s pitching staff is projecting a low OBPA (on base percentage against) while the Away team has an average pitching staff.  Which scenario is more favorable according to Log5?  The Away team with superior hitters and average pitchers, or the Home team with above average hitters and average pitchers?

If you replace BA from the above formula with OBP and PA with OBPA, you will calculate the following expected OBP for each team:

Away = .294 and Home = .320

Because the Home pitchers are extremely superior compared to the League Average, the Away team’s projected OBP according to Log5 is inferior to the Home team’s OBP.  Some would find this very surprising considering that the Away team’s OBP is superior to the Home team’s OBP and above the League Average!

It’s calculations like the Log5 formula that can be the starting point or core of any baseball system or simulation model.  It can even be extended further with the use of some additional advanced research.  You’ll notice the stated Log5 formula gives equal weight to the hitter’s batting average, pitcher’s batting average against, and the League Average.  Depending on the statistic, the importance of each variable can vary.  As it relates to OBP or batting average, which is more important when you apply the Log5 formula?  The hitters’s projected OBP or the pitchers’s OBPA, or are they equal? 

Good luck with your handicapping, and don't forget to visit SBR's live MLB odds page to find the best prices.