Once a Year Occurrences in MLB
I was thinking about home runs, as one does, and wondered what the standard for a great season is. A number of home runs that tends to occur only once a year on average seemed like a decent way to think about greatness. My gut said it'd probably be around 50.
From that, several other questions followed. How would those seasons be distributed across all seasons? How far back in time should I go? I bet a bunch would be during the steroid era. What about other eras? How much of an outlier are the really great seasons, like Bonds' 73 HR in 2001? What about other stats? Would be cool to evaluate those as well.
Methodology
We'll be taking a look at a set of hitting and pitching stats, first plotted by season (while grouping stat values per season in a tooltip for each plotted item) and then by stat value (while grouping seasons per stat value in a tooltip for each plotted item). We're looking at all seasons from 1901 (commonly considered the beginning of "modern" baseball) to 2023 — 123 seasons.
I used FanGraphs to pull the top 123 season performances from 1901–2023 for each stat, which usually resulted in a few more due to ties at the end of the range. Those values are plotted on each chart, and weighted by the number of occurances per year or value.
For the season-based charts, I added markers for MLB eras to add context, and for the stat-based charts, I added the min and max values at the ends of each row.
Hitting stats that occur once a season, on average (viewed by season)
A few things jump out here. It's immediately clear the impact changing the ball in the live ball era had, as we see a major increase in most hitting stats in 1920. We can also see the effect that integration had in leveling the playing field, which countinued through the expansion and free agency eras. The steroid era saw a massive boost in great seasons, followed by a massive rise in strikeouts in the post-steroid era. Lastly, it's cool to see stats like wRC+ and fWAR, which account for external factors like era and ballparks appear, to be pretty evenly distributed across eras.
Pitching stats that occur once a season, on average (viewed by season)
Not surprisingly, we see the opposite of a lot of the hitting stats, with heavy distributions of great seasons in traditional pitching stats during the dead ball era. However, as strikeouts become much more prevalent in recent years, we also see a lot of great performances beginning in the steroid era.
Hitting stats that occur once a season, on average (viewed by stat)
Viewing by stat instead of season helps answer the question the first 2 charts don't — how great are the great seasons? The best Triples season of 36 by Chief Wilson is miles ahead of the 2nd best season, for example. Also, Barry Bonds owning the top 3 Walks seasons by a massive margin also stands out.
Pitching stats that occur once a season, on average (viewed by stat)
First off, Pedro Martinez. Several of these stats just validate what is already well known — that he put up some bonkers numbers during an era of dominant offense. Also, Carlos Silva had 9 BBs total in 2005!
Final thoughts
This was a fun exercise, but I realize it's based off of a pretty simple and strict definition of "greatness". Many of the stats tend to bunch up in specific time periods, so if I were to expand on the concept I'd want to be able to easily filter ranges of years. Also, I'm using historical eras that are pretty commonly cited, but I found some interesting research challenging those assumptions (1, 2). Even just looking at the first chart, I'd argue that the steroid era lasted a little longer.