Quantifying Outlier Seasons

I’ve always been fascinated by the outlier season where a guy puts up numbers well above or below his career pattern (Mark Reynolds’ 2009 steals total is one of my favorite examples). I wanted to take a look at the biggest outlier seasons in baseball history. To do this, I ran the data on every player-season since 1950 and calculated a z-score for each season based on the player’s career mean and standard deviation for that stat (only including qualified seasons). While the results were interesting, in my first pass through I did not control for age and the results were largely what you would expect – lots of guys at the beginning or ends of their careers.

On my second pass, I rather arbitrarily restricted the age to 25-32 to attempt to get guys in the middles of their careers. I think these results ended up being pretty interesting. The full list is here, but I’ll highlight a few below:

I had never heard of Bert Campaneris, but it turns out he was a pretty good player who put up 45 career WAR, mostly as a speedy, light-hitting, great-fielding shortstop. But in 1970, he briefly turned into a power hitter. He hit 22 home runs, his only season in double digits. He hit two in 1969 and five in 1971, playing full seasons both years. So this wasn’t even a mini-plateau. This was a ridiculous peak that he would never come close to again. We don’t have the batted ball data to dig further, but I would love to know just what was going on that year.

Dawson, on the other hand, was a pretty good home run hitter who usually hit 20-30 a season, except in 1987 when he blasted 49. Usually guys hitting crazy amounts of home runs in the late 80s through the 90s wouldn’t be that interesting, but these guys played for a long time after, never coming close to their 1987 totals again.

The guys on the downside are all fantastic home run hitters. With guys playing a full season and falling this short of their numbers, it’s always a possibility that they were playing hurt. Schmidt did indeed play hurt in ’78, but a quick Google for Thomas and Carter brought up nothing, making it all the more inexplicable.

As I mentioned above, in 2009 Mark Reynolds went 44 HR/24 steals. That was Reynolds’ only season stealing more than 11, but it “only” registered a z-score of 2.0. The three guys listed here blow that out of the water. Zeile had his season early in his career so it could have been a case of a guy losing speed or getting caught too many times and then being told to stay put. But Palmeiro and Yaz did it right in the middle of their careers. Palmeiro’s stolen base record consists of usually stealing 3-7, and getting caught 3-5 times. But in 1993, he decided to steal 22 while only getting caught 3 times. The next year he was back to his plodding ways.

On the negative side, Crawford’s struggles have been well documented. Driven by a .289 OBP and possibly declining health, Crawford’s 18 steals in his dismal 2011 season were the lowest amount of his career in a qualified season by far. We knew it was a shocking performance at the time, but I didn’t fully grasp its historical significance.

The last things I will look at are plate discipline numbers. They differ from home runs and steals because they represent hundreds of interactions, thousands if you consider individual pitches, rather than the dozens that the former two represent.

Mantle’s 1957 season deserves some attention (although he put up 11.4 WAR so it probably gets plenty of attention). That year, he put up the second best walk rate and the best strikeout rate of his career, at age 25. After that he went right back to being the great player he was before, albeit with slightly worse plate discipline stats.

Except for Money who was a guy early in his career working his way into better walk rates, this is something I don’t have a great explanation for so I’d love to hear theories. Why did Ripken in 1988, right in the middle of his career, take a bunch of walks and then never do it again to that degree? Likewise, how was Brett Butler able to cut his strikeout rate from 8.7% to 6.3% in 1985 then jump back up to 8-10% for the rest of his career?

Before I corrected for age, I got a bunch of results of guys at the tail end of their careers doing what you would expect. I do want to highlight one of them, however. In 1971 at age 40, Willie Mays had a 3.7z walk rate and a 3.1z strikeout rate. He walked a ton, but also struck out a ton. Added with his 18 home runs, that season he had a robust 47% three true outcome percentage. As the z scores show, it was a radical shift from anything he had done in his career and impressively, he used this new approach to put up a 157 wRC+ and 5.9 WAR. Apparently that guy was pretty good.

This piece identifies the biggest outlier seasons in history, but is crucially missing the why. And unfortunately, for most of these that’s not something I have a great answer for. If you have enough player-seasons, you’re going to expect some 3z outcomes. But historical oddities are one of the joys of baseball and each of the 3z outcomes is the product of a radical departure in underlying performance. I think it would be fascinating to talk to some of these guys and see what they have to say about why things went so differently for one season.