Dead Ball

Handling HR for Dead Ball Era Players

Any sim aspiring to encompass the entirety of baseball history must make some philosophical commitments on how to handle Dead Ball players. In NSB this includes the seasons 1893 - 1920. While the end of the dead ball era is usually dated to 1919, home run percentages in 1920 remain low enough to keep the designation.

Throughout the life of our sim we have been tweaking the probabilities for dead-ballers, gradually adjusting them toward the modern era. In the early days of NSB, dead ball pitchers were among the most dominant pitchers in the system. In the summer of 2018, another in a series of adjustments was enacted that increased yet again the opponent HR tendencies. Let's break it down. But first, it's important to remember that there are myriad adjustments in any batter-pitcher matchup, and they are more complex than what we can cover here. We offer the following in order to demonstrate conceptually how we are handling these players.

In what follows “home run percentage” always refers to the percentage of hits that were homers (HR / HITS).

Adjustment 1

Because a great many homers of the era were of the inside-the-park variety, and because some old parks had no fences, it's very difficult to assume that a dead ball pitcher giving up few HRs actually has the ability to thwart HRs in a modern context. So, the first adjustment we make is to move the opponent HR percentage (Opponent Home Runs / Opponent Hits) toward the league norms for the season.

For example, in 1901 Christy Mathewson gave up 288 hits, only 3 of which where homers. His Opponent HR percentage is therefore .010 (3 / 288). The league percentage in 1901 was .022. So, in Adjustment 1 we double-weight the league opponent home run percentage, add it to Mathewson's personal opponent home run percentage, and divide by 3. Here are Mathewson's numbers plugged into the formula:

(.010 + (.022 * 2) ) / 3 = .018

The result, .018, now reflects a new (expected) home run percentage, with a good bit of the HR randomness eliminated. If we now multiply this new percentage (.018) by Mathewson's 288 hits, we get an adjusted opponent HR total of 5.18. This is a raw figure, prior to any other adjustments. It's purpose is simply to account for the largely random nature of home runs in the dead ball era. The 5.18 number will fluctuate up/down once we consider other adjustments for 1901, e.g., the effects of Mathewson's home park (in this case the Old Polo Grounds).

HR tendencies for batters between 1893 - 1920 are handled by the same method explained in Adjustment 1 above.

Adjustment 2

Adjustment 2 only comes into play when a dead-ball-era player (1893-1920) faces a modern-era player (1921-present). Let's suppose that Mathewson is facing Henry Aaron from from 1957. In 1957 the NL home run percentage was .106, meaning that 10.6% of all hits were HRs. Adjustment 2 halves the difference between the HR pct. in 1901 and 1957 and then uses that figure to nudge Mathewson away from his own league/season and toward the batter's league/season. Thus:

(.018 + .106) / 2 = .062

So, if we could imagine Mathewson playing in the 1957 NL, we would expect that .062, or 6.2% of his 288 hits, to be homers. Mathewson's baseline opponent home run projection now jumps from 5.16 to 17.86. Of course, Aaron's HR pct. (44 HR / 198 H) is more than double (by a factor of 2.09) his league in 1957. Thus If Aaron were to bat exclusively against Mathewson for an entire season we would expect him to hit 38+ home runs, more if ballpark conditions were favorable.

Now, since HR percentages are tethered to a pitcher's opponent batting average, playing Mathewson's 1901 season in a league stocked with above-average hitters and above average power, we wil likely find that Mathewson cannot get hitters out at the rate he accomplished in 1901 when he had an opponent batting average of .230. Let's imagine that his opponent average rises to .242. That would mean instead of yielding 288 hits for the season, Mathewson might now yield 303, raising his baseline HR expectation from 17.86 to 18.8. And if hitters have more than average power as we have stated, that number would rise again.

Again, Note that Adjustment 2 is unnecessary when a dead ball batter matches up against a dead ball pitcher. In those cases, the tendencies of the dead ball era are replicated exactly.

Now Let's Get Really Complicated

The simulator knows and uses stat splits, so at any given point in the calculation we often will be dealing with home/road, left/right, and bases empty/runners on breakouts. When we consider split stats together with ballpark adjustments, the relative strength of a pitcher's OAV to his own league, the relative strength of a batter's AVG versus his own league, weather conditions, and so on, the calculations are significantly more complex than the examples adduced above. Nevertheless, we hope this gives you some appreciation of the steps we take to handle HR tendencies for dead ball players. Other important stats like walks, strike outs, doubles, triples, and ground/fly tendencies don't have near the impact the home run, so these stats do not receive adjustments 1 and 2 as described above.

Can I See Some Basic Adjustments Before I Draft a Player?

Yes! In the database window, you will see 2 buttons ('H' and 'A') on the toolbar strip. These will toggle you between historical and adjusted stats. Adjusted stats are shown in red as indicated in the example below.

The adjusted numbers represent what you can expect WHEN AND IF Mathewson pitched in a “league of historical means” spanning the entirety of our system (1893-present) in a neutral park. This is a fiction, of course, since you are unlikely to play in a league of perfectly balanced historical means or averages. Yet it can give you some idea what to expect. You will also find diagnostic help by examining the Sim Drill Down Tab in the Database window where you can observe total (and split) accumulated simulation results.