Paul Articulates – Too much of a good thing

 I will preface this article by saying that I believe in analytics – I have both Bachelor’s and Master’s degrees in engineering, so math has been the most important tool in my box.

However, I also believe that you can have too much of a good thing, and the 2021 Mets have been gluttons at the analytics table.  Their post-season aspirations are suffering from gout.

All baseball analytics boil down to probability and statistics.  Crunch a whole lot of data on many aspects of player and team performance, and then analyze that data to predict outcomes.  When an outcome is predicted, a manager can use tactical measures on the field to manipulate that outcome.  Also, in the cage or the bullpen, a coach can adjust player mechanics to achieve a desired outcome. 

There are many examples of this theory in practice in today’s Major League Baseball, such as:

-          Defensive shifts to cover areas of the field where batters are most likely to hit

-          Substitutions to create lefty-righty or righty-lefty matchups

-          Removal of pitchers before facing a lineup the third time around

-          Retooling of a batter’s swing to increase launch angle

-          Throwing more pitches outside the strike zone to players with high chase rates

Wow!  The use of science and math to gain a tactical advantage!  How could you go wrong? 

The answer lies in not understanding the limitations of the theory.   Probability theory is based on something called the “law of large numbers”.  From Wikipedia, “According to the law, the average of the results obtained from a large number of trials should be close to the expected value and will tend to become closer to the expected value as more trials are performed.”  These “trials” are expected to be identical events performed many times in the same environment.  Think of a pair of dice rolled over and over on the same table.  The probability of rolling “7” is 17%.  Now let’s apply that to baseball.  There are certainly a large number of events since there are 30 teams playing 162 games per year with rosters of 26 players.  There have been almost 150,000 at-bats in 2021 so far, so that’s a large number.  But wait, part of the criteria is that they have to be identical events performed in the same environment.  An at-bat against a tired rookie left-handed pitcher during the daytime in Colorado is not the same as an at-bat at night against Gerrit Cole in front of 50,000 fans in Yankee Stadium.

So the data is not clean, and the probabilities derived from it are not accurate.  There are still valid indicators that come from it, but there are several ways I believe the Mets are misusing the indicators:

Case 1: Pitching changes.  Luis Rojas consistently removes starting pitchers by the fifth or sixth inning, even with pitch counts that would reasonably allow for an inning or two more without health impact.  Is this because the analytics say that batting average against goes up on the third time around?  Is it to leverage an advantageous righty-lefty matchup?  We don’t know and he won’t tell, but the effect is that the Mets then have to fill 4 innings of work from their bullpen night after night.  Add to that the fact that Mets relievers (almost) never go more than 1 inning, and over a few months, the bullpen gets fatigued.  (By the way, there are no splits in the data for fatigued vs fresh pitchers).   I would propose that this misused analytic leads to additional runs allowed in the later innings.

Case 2: Launch angle.  It is no secret that many of today’s hitters are having their swings adjusted to pull the ball with optimized launch angle for home runs.  For certain power hitters (e.g. Alonso, Conforto) this can improve run production.  But for others it can have disastrous effects.  Jeff McNeil, a lifetime .319 hitter coming into the 2021 season is suddenly struggling to get to .250.  With his world-class bat-to-ball skills he should be having multiple hit games against every team that employs a shift against him, but instead he has struck out or rolled over into 4-3 outs repeatedly.

Case 3: Over-use of pinch hitters.  The use of a pinch hitter in a critical situation can be a manager’s greatest contribution to team success.  However, I think the Mets overuse the lefty-righty splits every ballgame.  The same manager that can’t justify leaving his starter in for a third time around the lineup is bringing in 3-4 pinch hitters per game – meaning that he is removing a guy that has seen the opposing starter a few times already.  It’s brilliant when a lefty pinch hitter forces the other team to remove their righty starter that was dominating for 7 innings.  It’s dumb when Nido pinch hits for McCann (they’re both righties batting around .230) because his stats are better against the opposing pitcher over 20 career at-bats.

Case 4: The over-shift.  The defensive shift works well when the opposing batter hits the ball as expected and won’t adjust his swing knowing there is an open field.  Even though this still happens in baseball, I have seen the Mets defense shift three players to the right of second base with a runner on first.  It has disrupted double plays and left huge open gaps for opposing hitters to exploit.  With the outstanding fielding range of both Lindor and Baez, it is a sin to bunch them up together on the same side of the infield.

I could go on, but you get the point.  The Mets’ 2021 season has been a confounding series of disappointments.  It has not been a player or two that has underperformed – it has been a majority of players on a very capable roster.  I believe that the common factor is the over-use of analytics and I look forward to the time when this team finds a reasonable balance.