Does Spring Training Matter?

It Happens Every Spring

“Hope Springs Eternal”, from the poem by Alexander Pope, wasn’t written about sports, but it has been co-opted because it adequately captures the feelings of optimism, warmer weather and better times ahead for fans of MLB.

Major League baseball’s statistics are rich in quantity and history. But Spring Training has changed significantly. Gone are the days when players needed time to adjust to major league caliber after their real jobs (including military service). Today’s major-leaguers are in top-notch condition year-round, the result of advancements in training and wanting to protect massive performance-based contracts.

So does a player’s performance in the Spring make a difference, or is it just about feeding the Hope?

Data Enrichment

The object of this analysis was to determine if a player’s spring training batting average (base hits / at-bats) was useful in predicting his regular season average. But some players aren’t kept around for their batting prowess (i.e. pitchers), and still others aren’t kept around at all (i.e. minor leaguers getting a chance to make a very selective roster).

In order to limit the analysis to players with significant appearances in both situations, minimum at-bats during Spring Training (20 AB) and the regular season (150 AB) were enforced.

Another situation that needed to be addressed (for the team analysis) was the trade.  Some players shuffle between teams during the season. In the event of a trade, the player’s total average was used, but the team that he finished Spring Training with was used for labeling.

The total number of players meeting these criteria was 352.

Batting Average: Spring Training vs. MLB

The scatterplot below shows the 2014 Spring & Regular Season batting averages for players meeting the entrance requirements.

Baseball and data science - MLB 2015

 (Click to enlarge) 

Correlation= 0.206.

Slope of trend line = 0.0908 (p=0.0001).

The slope suggests that a higher Spring Training average will result in a higher regular season average. However, the effect, while statistically significant, is very small. For instance, a 100-point average increase in Spring Training BA would only result in a 10-point regular season boost.

The graph was made with the Tableau desktop version, which is a great tool for visualizing many types of data.

Analysis

Players get fewer at-bats in the shortened Spring season. As a result, the spread of averages tends to be much higher. Results at the extremes tend to revert to the league average, since those aren’t sustainable through a 162-game season (Michael Brantley, who finished with a robust 0.500 BA for Cleveland, had to settle for a still-impressive regular season BA of 0.327). There hasn’t been a regular season average above 0.400 since 1941. And players who bat extremely low (below 0.200) may not last long in the big leagues.

Of the qualifying players, the five with the worst batting average decrease between Spring Training and the regular season appear below:

Table 1 - Players with Worst BA Drop in 2014

batting average data 2014

The players appearing on this list have the distinction of having the biggest performance drop from Spring Training in 2014. Brad Miller of the Mariners experienced the worst drop at 189 points.  Any of these players would’ve gladly given up their lofty spring performance for more tangible results. The table below shows the players with the most improvement after Spring Training.

Table 2 - Players with the Best BA Improvement in 2014

batting average data improvement

Each of these players had a terrible Spring Training batting average (implying they had the most room for improvement). Josh Willingham of the Twins experienced a 147-point increase, and still managed only 0.215 in the regular season (a spring training BA of 0.068 allows for a great deal of upside).

All of the code and data to reproduce these analyses, or do your own, is available at the following repository: https://github.com/joebluems/SpringTraining.git

Conclusions

A player’s performance in Spring Training is slightly indicative of the season to come.  The Spring season can be thought of as a preview or sample of the main event. As with most samples, size is important; an impressively high spring training batting average usually implies only that what is up must come down.

So while it’s been shown that ballplayers do still use the spring to warm up, a poor performance isn’t exactly the end of the world. In other words, it’s no reason to give up hope.

Streaming Data Architecture:

New Designs Using Apache Kafka and MapR Streams

 

 

 

Download for free