Sunday, August 18, 2013


The news reports will tell you that the Los Angeles Dodgers' 42-8 record from June 22nd through August 17th is the best such 50-game streak since 1942. And that's definitely good to know, in a sound bite kind of way.

Less frequently those reports will provide the factoid that the greatest 50-game streak in baseball history (defined as 1901 to the present...) belongs to the Chicago Cubs, which occurred in the deadball era. The 1906 squad, which finished the year 116-36, had a 50-game stretch in the second half that year where they went an incredible 45-5.

What we're not getting from any source, however, is just how often baseball teams have relatively protracted periods of playing close to .800 ball.

For the sake of what we're doing here, we're going to term such an occurrence as a "scald," and will define it as being any fifty-game stretch in which a team wins 39 or more games. (There is one exception to the 39-game rule: teams who win less than 39 games but have higher than a .780 WPCT due to tie games, as you'll see in the 1914 Boston Braves below.)

We will only count one such "scald" per team in the year during which it occurs, rather than counting up the number of equivalent W-L records (or lower: if a team goes 41-9 in games 76-125, then goes 40-10 in games 77-126, we're not going to include that one as well). We want to get the most elemental list with which to work.

When we do that, we find that there have been 53 such "scalds." The Yankees have had nine of these (1927, 1928, 1938, 1939, 1941, 1947, 1953, 1978, and 1998). The Red Sox have had four (1912, 1915, 1946, and 1978). The A's (who had three during the last decade) have had eight: five in Philadelphia (1902, 1913, 1914, 1929, 1931) and three in Oakland (2001, 2002, 2005). The Giants have had seven, all in New York (1904, 1911, 1912, 1913, 1936, 1951, 1954). They are the only team in baseball history to have "scalds" in three consecutive seasons.  The Cubs have had five (1906, 1907, 1909, 1910, 1945). Four of theirs came in a very productive five-year period.

The only times there have been "scalds" by teams in the same league during the same year: 2001 (A's and Mariners); 1977 (Yankees and Royals); and 1909 (Cubs and Pirates). Just three times in 112 years.

Of course, in today's game we can have teams from different leagues who each had "scalds" with neither of them making it into the World Series: 2002 (A's and Braves).

The Dodgers have just achieved their second "scald." The first one came in 1953: this was the Brooklyn team that won the most games ever in franchise history (105).

Out of the 52 scalds that have fully completed seasons, we find that 44 of these teams have either won pennants or finished first in their division. Obviously, there aren't very many "fluke" teams to be found on these lists.

We've broken these out into three separate lists--deadball era (1901-19), liveball era prior to expansion (1920-60), and expansion era (1961- ). As the charts will show, there was a 21-year hiatus between "scalds" between 1954 and 1975.

We've included runs scored/runs allowed data with the teams' records, as well as two other measures that derive from that info: the Pythagorean Winning Percentage (PWP) and the difference in projected wins (PW) between the PWP and each team's actual winning percentage.

For the 53 teams with "scalds," it turns out that they have averaged 3 more wins in real life than what is projected for them by their PWP.

The most basic thing we can see from that data is that certain teams have had a sizable "luck" quotient during their "scalds." The Dodgers' just-completed skein is one of those where the team won at least five additional games more than what their PWP projected to be the case. There have been eight teams whose "scalds" have involved at least five additional real-life wins as compared to their PWPs: the 1907 Cubs, the 1909 Pirates, the 1913 A's, the 1923 Reds (one of the few non-pennant winning teams to make this list), the 1930 Cardinals, the 1951 Indians (another of the non-pennant winners), the 1983 White Sox, the 2001 Mariners, and now the 2013 Dodgers.

When we compare the summary data of the "scalds" from these three eras, we find some interesting trends that might well have bearing on some other recent phenomena in what we might term (for lack of anything better...) the "applied theory of winning."

These trends are consistently going in one direction for each measure, from the deadball era through the  pre-expansion liveball era into our ongoing post-expansion 30-team model.

Teams who have scald are having a lower PWP as time goes on (.753 in the deadball era; .738 in the pre-expansion liveball era; .724 in the post-expansion era). They also have a lower real-life WPCT, but that's declined by just ten points instead of thirty.

The percentage of teams who "scald" who have less than a .750 PWP has increased from 59% in the deadball era to 67% in the pre-expansion liveball era. to 87% in the post-expansion era.

The percentage of teams who beat their PWP expectation by less than two games has dropped from 42% in the deadball era to 19% in the pre-expansion liveball era, to 13% in the post-expansion era.

The average gain in real-life wins over the PWP expectation has risen from 2.3 games in the deadball era to 3.2% in the pre-expansion liveball era to 3.6% in the post-expansion era.

Even though this is a small sample of teams, these trends track with the increasing standard deviation in real-life WPCT/PWP that has occurred in the recent past, where changes in the post-season and the ability to exploit structural features of the expanded size of the pitching staff have come into play.

Increasingly, this "optimize to exceed expectation" is moving from anomaly to nascent strategy. The downturn in offense in the past few years seems to driving a good bit of what's occurring, and the little flurry of teams who markedly exceed their PWP projections may turn out to be more than just a random event.

As we say here--stay tuned.