Glenn DuPaul recently wrote a very interesting article here on Beyond the Boxscore, where he looked at Albert Pujols and his declining walk rate. Reading the article sparked my memory of two other recent players that I recalled had decreased walk rates in National League to American League transitions: Adrian Gonzalez and Prince Fielder. I began to wonder whether switching leagues led to a measurable change in walk rate for the average hitter.
For the purposes of this investigation, I will use the time period from 2002-2012, giving me ten year-over-year change sets for analysis. I chose to look only at players who achieved at least 200 plate appearances in each year. I wanted to make sure that BB% had stabilized in each year of the study, and 200 PA is the most conservative of these two studies regarding metric stabilization. Trying to look at players who changed leagues during a season would add some complexity, in particular when they did not accumulate more than 200 PA on one or both of the teams in that season. To make things simple, I decided to only look at players who changed leagues during the offseason. In other words, players who played all of year N in one league, and all of year N+1 in the other. I also excluded those players who missed the entire N+1 year (due to injury, for example) and changed leagues in year N+2. The sample set of players is thus fairly consistent in the manner in which they changed leagues.
Over the decade in question, walk rates in general have been falling slightly on average in both leagues. Since there are no pitchers within the sample set of batters used for this study, we will consider only non-pitchers in the control group. In the American League, walk rates of non-pitchers have fallen on average 0.04% per season. In the National League, walk rates have declined on average 0.11% per year.
In addition, on average, walk rates in the National League have tended to be robustly higher than walk rates in the American League. Over the period used in this study, the National League walk rate was typically 0.50% higher than the American League walk rate when considering only non-pitchers.
With these two known controls, I can adjust any walk rate deltas to remove any expected change in walk rate just from typical year-to-year and league-to-league movement.
BB% Change Results
|League Switch||Avg Y-Y BB% Delta||Avg League Y-Y BB% Delta||Avg New League BB% Delta||Adjusted Y-Y BB% Delta||Sample Size|
|AL to NL||0.56%||-0.11%||0.50%||0.17%||82|
|NL to AL||-0.97%||-0.04%||-0.50%||-0.43%||92|
|AL to AL||-0.29%||-0.04%||-0.25%||76|
|NL to NL||-0.15%||-0.11%||-0.04%||100|
Data: Fangraphs, 2002-2012, hitters with >= 200 PA
The table is binned into the four possibilities for league-to-league movement given that a hitter is changing teams. The "Avg Y-Y BB% Delta" column indicates the actual average walk rate differences for hitters changing teams in the offseason. These numbers are then adjusted to account for normal expected walk rate changes due to league-wide year-to-year decline as well as a change of league, where applicable. The adjusted values are then listed in the "Adjusted Y-Y BB% Delta" column. The table also lists the sample size for number of players in each grouping over this time period.
The results indicate that after accounting for expected league-change-induced walk rate differences, hitters changing leagues do experience a higher magnitude shift in their walk rates.
Hitters moving from the American League to the National League see an average walk rate increase of 0.17%. Conversely, hitters transitioning from the National League to the American League see an average walk rate decrease of 0.43%. This direction of walk rate change aligns with the example data points of Pujols, Gonzalez and Fielder that ignited this study.
Interestingly, players that switch teams within the American League also have tended to walk less in the first year on their new teams, but only by 0.25%. Players moving within the National League have experienced almost no difference in walk rate, with a mere 0.04% decline year-over-year.
Are there confounding factors in this study? Yes. For one, by restricting the sample only to players who change teams in the offseason, I suspect that free agents will make up a reasonable portion of the sample. Free agents by definition will have had to be playing in the league for a number of years, so I suspect my sample of players would be older than the league average. In fact, that is true, as the average age of all players who changed leagues in this study was 30.3 in Year N to 31.3 in Year N+1. The average age of baseball players over this time period was 29.0. Do these two years make a difference? Well, thanks to Bill Petti we know that walk rate is one of the most stable hitting metrics, year-over-year. We also know though that plate discipline tends to peak in the early thirties, or around age 32 to be precise. Since plate discipline stats do make an impact on walk rate, we could perhaps expect players to walk more than average in this group. If anything, this serves to diminish the AL to NL difference slightly, but highlight the NL to AL delta found in this study.
Another major factor not separated in the table above is the fact that walk rate includes both non-intentional and intentional walks. Jeff Zimmerman pointed out this fact in Glenn's article, and certainly changes in intentional walk rate could account for a great deal of the differences noted in the above table. Given this issue, I performed the same study, only this time with intentional walks removed, to isolate non-intentional walk rate changes.
In considering purely non-intentional walks, there has still been a gradual decline in rate year-to-year in each league during this time period. The American League as a whole has averaged a 0.03% decline, while the National League has typically dropped 0.07% per season.
After extracting intentional walks from the picture, the average walk rates from the National League and American League move closer together, with the NL still 0.22% higher than the AL.
Once again by knowing these two controls, I can adjust any non-intentional walk rate deltas to remove any expected change in the rate just from typical year-to-year and league-to-league movement.
NIBB% Change Results
|League Switch||Avg Y-Y NIBB% Delta||Avg League Y-Y NIBB% Delta||Avg New League NIBB% Delta||Adjusted Y-Y NIBB% Delta||Sample Size|
|AL to NL||0.37%||-0.07%||0.22%||0.22%||82|
|NL to AL||-0.63%||-0.03%||-0.22%||-0.38%||92|
|AL to AL||-0.21%||-0.03%||-0.18%||76|
|NL to NL||-0.06%||-0.07%||0.01%||100|
Data: Fangraphs, 2002-2012, hitters with >= 200 PA
With intentional walks removed and deltas adjusted for league averages, the results are still actually very similar to what we saw in looking at general walk rate. Players switching teams and ending up in the National League had slightly higher levels than the original adjusted deltas, with a 0.22% increase for those moving from the American League, and now an almost flat 0.01% increase when changing within the National League.
The deltas for players moving to American League teams get slightly suppressed from their total walk rate levels. Players shifting from the National League experienced a 0.38% decrease in non-intentional walk frequency, while those changing teams within the American League posted an average decline of 0.18%.
The magnitude of the deltas may not be overwhelming, but a trend is certainly visible. Hitters tend to find it easier to draw walks when moving to a National League team, but tougher to induce bases on balls when shifting to an American League team. As for reasons for this trend, I would suggest that the American League has been a stronger league for most if not all of the period used for this study, as indicated at least partly by the relative records of the two leagues in inter-league play. Stronger in this sense could mean better pitching, or more complete offensive units that make it harder for pitchers to pitch around hitters to get to face weaker batters in the lineup.
As for the three sluggers who were mentioned off the top who recently changed from the National League to the American League, all three experienced drops in walk rate even greater than the averages found in this study. The average player changing leagues in this direction had a 0.97% decline in overall walk rate and 0.63% decline in non-intentional walk rate. Adrian Gonzalez saw his overall and non-intentional walk rates fall 3.1% and 0.8%, respectively. For Prince Fielder, those same drops were 3.2% and 1.1%. For Albert Pujols, the declines were 1.6% and 1.7%.
One other topical case to watch for in 2013 is new Toronto Blue Jays outfielder Melky Cabrera. This will mark the fourth consecutive year that Cabrera will have switched leagues in the offseason. His previous three transitions have all followed the trend discovered in this study: both his overall and non-intentional walk rates have climbed in both occasions where he moved to the National League, and both fell when he returned to the American League in 2011. Not to say drawing walks is particularly a strong suit in Cabrera's game, but it would seem that on average we should expect him to walk less in Toronto than he did in San Francisco.
In the next article, I plan to look at the effect on strikeout rate when hitters change leagues.
You can follow me on Twitter at @MLBPlayerAnalys. Follow @MLBPlayerAnalys
Credit and thanks to Fangraphs for data upon which this analysis was based.