Making Sense of MLB Stats: Correlation of Hitting Statistics and Fantasy Points

Making Sense of MLB Stats: Correlation of Hitting Statistics and Fantasy Points

Welcome to the third installment of Making Sense of MLB Stats. In this article we will discuss the correlation between specific player batting stats we can research and how strongly they predict a factor we are interested in.

Before we get started, I will share a brief overview on understanding correlations. All correlation results will return a value between 1.0 and negative infinity. A 1.0 correlation means that in the specific relationship being compared, as value X increases as does value Y – there is no stronger correlation than a 1.0 correlation. This will be a rare phenomena for us to experience in DFS, however, we can still look at the correlation coefficient in the tables below to understand how strongly related a stat is to the desired result.

As a general rule of thumb, see below table that explains the level of relationship versus the correlation coefficient.

For our purpose, we will be targeting all statistics that have a correlation greater than .5 with .5 being the absolute baseline of if that statistic is worth us using.

Fantasy Points Correlation

We will get right into the meat of this article and discuss each statistic and its relevance to fantasy points per season.

It should come as no surprise to see PA lead to the most fantasy points scored per season. The more opportunities a player gets, the better chance he has to score fantasy points. We discussed this in a previous Making Sense Of MLB Stats article titled “Impact of Lineup Slot."

The next few statistics were the most interesting to me. wRC+ and wOBA came in near identical with their predictive value to future fantasy success. This means that when you are looking to increase the amount of fantasy points for your team, the foundation of your research should be surrounded by these two statistics.

When we look at the next tier, I was completely shocked to see that SLG, OBP, AVG all have a higher correlated value to fantasy points than ISO. Remember, we are directly comparing the impact of a statistic versus total fantasy points. This means that when we are building cash game lineups, we should be less concerned with ISO and more concerned with wOBA, wRC+ and SLG.

When we think about this, it actually makes perfect sense. ISO is a power statistic and with power comes a higher strikeout rate. This means fewer times on base and less opportunities to score runs and tally RBI. I will touch on the benefit of ISO later in this article.

The two weakest statistics on future fantasy success is BB% and K%. Neither of these stats can truly help you predict a high fantasy output.

HR Correlation

While building cash games, we want high floors of players who score fantasy points on a more consistent basis, however, in GPPs we want to target players who have that high upside and home run potential.

As displayed in the above graph, no statistic correlates home run and power upside better than ISO. Considering ISO stands for isolated power, this should come as no surprise. When building GPP lineups and a punt play is necessary, taking the player with the higher ISO is a proven method to increase your potential of a home run in your lineup.

Extra-Base Hit Correlation

Targeting players who hit for extra-base hits is a fairly even split across wRC+, wOBA and ISO. By targeting these three stats you will not see much of a difference in predictive value in the short term.

Run and RBI Correlation

I have paired these two together to show that there is a larger correlation to predicting RBI over runs based on the statistics provided. It is also to be noted that each statistic has a similar strength of correlation and can each be used in your research of predicting RBI.

When we look at runs, we notice that ISO is significantly less than wRC+ and wOBA. This makes sense as well because ISO is a power stat while wOBA and wRC+ are more of an “on-base" stat.

In closing, you will want to build your lineups based on the correlated statistic to the game you are playing. If you are playing a GPP game, the stats you should be interested in better be different than if you are playing a cash game.