In the past 30 years, baseball fans (and increasingly baseball managers) have used statistical methods to validate or invalidate a baseball 
manager's strategies. Most of these strategies are designed to either create runs (if the manager's team is batting), or prevent runs (if the 
manager's team is in the field). Since these strategies are based on the batter's or pitcher's previous records, most use linear regression to 
see how the previous record predicts the outcome. 



        We focus on two main strategies: using the handedness of the opponent to select the player and using the record of the player against 
the current opponent to select the player. Our analysis is based on a newer statistic, called runs created. We evaluate the use of linear 
regression on this context. In particular, we found that although we were using a standardized statistic that directly reflected the run 
contributions of the players, these statistics were not very useful as predictors in a linear regression, and perhaps other methods of 
validation should be used.