Saturday, January 28, 2006

A Word About Historical Statistics

A Word About Historical Statistics

Attached is a box score of a game between Hollywood and Sacramento, played on 4/28/1926. You will note that the Stars' Charley Gooch hit a homer in this game. However, according to the Old Time Player Database and your Cyclopedia, he is not credited with any homers in 1926.

Isi Baly

Isi seems to chide me about missing the Gooch home run. But if you check the guides, Charley Gooch is listed with “0” home runs in the guide. I did not go back and recompile every Coast League season from 1903 through 1957. As I explained to Isi,, I correct the mistakes as I come by them, just like the editors of the encyclopedias do. I did, however, recompile season batting records 1903-1906, and seasonal pitching records 1903-1910. I also attempted to compiled records for all the players who did not appear in the guides— for the most part “less-thans”— though I probably missed a few over the1906-1937 period, because the only way to really make sure that you have everybody is to compile the complete season from box scores. Obviously, something that would be a “life’s work.” I also compiled individual pitching stats for every season that didn’t exist; e. g., saves.

For 1903, I found one player, Deacon Van Buren (who was listed many places as the league leader in batting), whose BA dropped from .360 to .319. Generally, most players were close to what the guides had, though far enough off to require a redoing of the stats.

What do I attribute the discrepancies to? I attribute the differences to sloppy record keeping. And pitching, over the years, was much worse than batting.

Did it get better over time? It tended to fluctuate: 1905, for instance, I found— after a couple of months’ work— to be almost identical to the official averages. Now that was a big disappointment. Though pitching records were very incomplete, so it wasn’t all wasted effort.

Stats for 1918 turned out to be spot on, though no pitching averages were published, so my work was not in vain.

The stats for 1930 were so bad that the Coast League brought in Irving Howe (who would later create Howe News Bureau)to do the 1931 stats for the league. (I plan on redoing the stats for 1930 whenever I get a chance. They must be atrocious for the PCL— never the league to care about the historical record— to declare them so.)

Willie Runquist started recompiling stats from 1938 on, and to give you some idea of what he found, I’ll quote from his 1938 Pacific Coast League Almanac: “Finally we would note that most of the differences may be so small as to be unimportant.”

But just as we begin to feel confident with the official averages in the league, they get worse in 1939: “The discrepancies are considerably larger than they were in 1938…”

What does this all mean? It means that for the Coast League statistics were generate in-house, and the quality depended on the person compiling them. And, generally, pitching records were worse than batting records. Therefore, recompiling pitching stats should be the priority, though several seasons, like 1930, should be done top to bottom.


Post a Comment

<< Home