When Triples Were As Common As Home Runs

The 1930s

When Triples Were As Common As Home Runs

After looking closely at National Pastime the other day, I’m convinced that Clifford Van Beek didn’t know how many triples each player actually hit in the 1930 season.

In fact, the more I look at the distribution of extra base hits, the more interesting this question becomes.

I did a real quick and dirty study using Baseball Reference data of the distribution of extra base hits over time. This is what the most recent few decades of National League history looks like:

The percentage of extra base hits that wind up triples has decreased steadily since 1991. This is presumably because of the decline of artificial turf stadiums, as well as the increase in stadiums with short fences. Notice as well that the raw number of extra base hits has increased — largely because of the crazy increase in home runs we’ve seen over this time. Of course, expansion played a role as well.

In contrast, take a look at the earliest years of the National League:

The raw number of extra base hits declined with the deadball era — but you probably already know that. Interestingly, the percentage of extra base hits hit for home runs increased in the 1890s, which I’d wager most people don’t know, only to fall again with the coming of the deadball era. The fact that there were only 96 home runs in the National League in 1902 is mind boggling.

And now let’s look at the 1930s:

The share of triples outweighed the share of home runs in the National League up until 1928, when the Ruthian method of hitting finally caught on in the other league. And then we had a really odd period of time between 1928 and 1933 when there were just about as many triples as home runs.

The crazy thing is that the same pattern exists in the American League. Here are the more modern seasons:

Remember that the National League had more artificial turf parks in the 1980s and early 1990s than the American League.

Here is the deadball era:

It’s very likely that all the good power hitters left the National League and went over to the American League in 1902.

And here is the 1930s:

Note that there were almost exactly the same number of triples and home runs in 1929, 1930, and 1931 in the American League. After that, home runs gradually gained the advantage, though there were still a considerable number of triples hit into the early 1940s.

So what does all of this mean? Well, for one thing, it means that the National Pastime cards almost certainly don’t have enough triples. You don’t see players with multiples of play result number 2 or 3, for example. Something odd is going on — and we’re going to take a look at some individual players to try to figure this out.

It also means that you need to keep this in mind when you do cross-era normalization. One problem that automated algorithms have is that they don’t usually convert one time of extra base hit into another.

I did a short study a while ago using OOTP. I took Barry Bonds out of his era and stuck him into the deadball era to see what would happen. You can check the video out here:

Now, neither you nor I know precisely what would have happen if this had really been the case. Bonds wouldn’t have been able to juice up in the early 1900s the way he could in the 1980s. Bonds would have faced a huge amount of discrimination regardless of his physical talents — and even if you ignore the racial issues, it’s not clear that Bonds would have favored a long ball game.

Based on the very quick above study, I would argue that you’d need to turn Bonds’ home runs into other extra base hits. In other words, Barry Bonds is still going to hit long line drives that fielders can’t catch. In those old ballparks, however, those are going to turn into triples, and perhaps occasionally doubles.

But that’s not what OOTP does. In fact, that’s also not what Diamond Mind Baseball does, or what Action PC Baseball does. The algorithms will simply turn those home runs into outs, which creates major cross era play distortions.

We’ll talk more about this soon in the context of OFAS type sets.