The K Zone

December 5, 2017

by Ian Joffe


Last year’s trade of Yu Darvish to the Dodgers excited not only Dodger fans, but fans around baseball, in a manner that reminded me of the excitement surrounding Max Scherzer’s pact with the Nationals a few years ago, and many other instances. In both these cases, an incredibly talented baseball player, who had already proven themselves, was moving from the American League to the National League. Supposedly, when a pitcher moves from the AL to the NL, their numbers should improve noticeably, mostly because they get a free strikeout every ninth at bat, but also because of what some see as a difference in playing style between the two leagues. At some points I try to take advantage of others’ notions of league difference. Last summer, my fantasy team swung a deal trading Darvish off my team for Luis Severino and Jean Segura (to fill a need at shortstop). Yet, I too am not immune to the league switch hype. Back in 2015, I put all my eggs in one basket, drafting Scherzer in the second round of my fantasy draft and not taking another pitcher until my entire lineup was filled. Darvish was alight as a Dodger, but nothing like what was expected. Minus a couple of absolute blowouts, he had similar statistics in both leagues, and is blamed by many for costing L.A. the World Series.  Unlike in the Darvish case, Scherzer did undergo small but significant and very consistent improvements after changing leagues. Between my two personal experiences, some would call me lucky in choosing the right guy to get better after changing leagues, and they would be right. Over the years, my friends and I have invested a lot in the potential for improved statistics, but none of us have ever actually seen mass evidence that players improve after changing leagues, something more than the anecdote of Max Scherzer, which can always be countered with the mediocre performance/total bust (depending on who you ask) of Yu Darvish. So, as I like to do, I set out to create that evidence, or disprove the theory entirely.

I started out with a data set of the 45,000 pitchers ever to play the game. I was quickly able to narrow this down. My first step was to eliminate all players from before the year 2000 in order to make sure all rules and most playing styles are constant with today. I then eliminated players that did not participate in both the AL and the NL. By sticking to pitchers that threw in both leagues, I eliminate the potential that one league just drafts better pitchers, and one league better hitters. Next, I trimmed off everything but a player’s first six years of service time (and then again eliminated players who did not play in both leagues during this time) to ensure that the league change, not the aging curve, accounted for any difference in the player’s performance over time. Finally, players whose samples were too small were cast out of the spreadsheet, and the comparisons could begin. Each player’s average statistics in the NL and AL was calculated, and from that, each league’s average statistic was calculated.

Some statistics stayed relatively similar between leagues. When a pitcher switches from the AL to the NL, their walk rate only goes up from 3.10 to 3.13 (a 0.9% increase). While it’s arguable that a pitcher should walk fewer players in the NL because the pitcher walks less often, this probably accounts for little difference (it’s not like the typical AL nine-holer is an on-base machine). Rather, I would say that this shows that there is little difference in the playing style between the two leagues in terms of pitcher strategy. American League managers do not pitch around more often than in the senior circuit. Instead of proving less of a difference between the leagues, the change in walks actually further legitimatizes the other statistical differences as pure functions of league difference, rather than strategy or talent level. The second smallest change occurred in home runs. The average AL pitcher gave up 2.09 home runs per 9 innings, while the typical national leaguer forfeited 2.14 dingers (a 2.3% change). This is change is significant, but only slightly significant. It would not account for any real difference in overall pitching. But, the last of the three true outcomes, strikeouts, shows a very significant difference between leagues. After moving to the NL, pitchers’ K/9 ratios move from 6.25 to 6.79 (8.6%). We know from the stagnancy of walk rate that this difference is for real, but what aspect of the league change accounts for that change? The answer lies in the change in opponent batting average, which goes from .263 to .275 (4.5%) upon a change in leagues. In the same time period that I used to sort through the pitcher stats, the average 9th hitter in the AL held a batting average of .241, and the average pitcher in the NL batted .138. The difference between those numbers is 42%, and 42% divided by the nine batters in the lineup makes 4.7%, almost identical to the 4.5% change in batting average from all over the lineup between leagues (perhaps pinch hitters for pitchers and their .225 BA accounting for the very small difference). So, what we find is that there is significant improvement moving from the AL to the NL (or significant decline moving from NL to AL), and that this change is entirely due to the difference in rules in terms of pitchers hitting. This change is summed up with earned run average. The pitchers in my data had an AL ERA of 4.74 and an NL ERA of 4.34 (9.2%), accounting for a change in 65 runs over a 162-game period: the pitching difference between the 2017 Astros and Giants, Cardinals or Phillies. This time, the old myth turned out to be very true: purely because of pitchers hitting, a pitcher becomes significantly better when they move to the National League. The Max Scherzers of the baseball show the true trend, while the Yu Darivshs are just tipping their pitches.