For those of you interested in the "what-ifs" of baseball history, presented here in several installments is an alternate history of the major leagues stemming from a point of divergence in 1892. When the American Association collapsed after the 1891 season, the National League absorbed four AA castoffs to become a 12-team circuit. In our timeline, 1892 bore witness to an experimental split-season setup in which the team with the best record in the first half of the season played the best team in the second half in a postseason series. Unfortunately, this arrangement permitted the winner of the first half to slack off for the rest of the season, making a farce of the second half pennant race. The split season concept did not go over well with fans and was dropped for the next season.
An Alternate History
In the alternate timeline (ATL) I provide here, the National League instead experimented with a different novel setup in 1892 than in our timeline (OTL). The unwieldy complement of twelve teams was split geographically into two more manageable 6-team divisions, an East Division containing Baltimore, Boston, Brooklyn, New York, Philadelphia, and Washington and a West Division containing Chicago, Cincinnati, Cleveland, Louisville, Pittsburgh, and St. Louis. These teams were already grouped together for scheduling purposes, so the geographic split seemed to follow naturally from this. Unlike in previous seasons, the schedule was unbalanced so that competitors within a team's own division were played more often than those in the other division. Not only did this alignment reduce travel costs, but it also provided for more exciting pennant (or division championship) races, rather than forcing teams bear the stigma of languishing in, say, 9th or 12th place in the standings for most of the season.
Beginning in 1892, the best team in the East Division would play the best team in the West Division in the "National League Championship Series" at season's end. This format would prove successful during the rest of the 1890s, and was adopted by the upstart American League in 1901. Also in this year, the AL and NL would pit their league champions against one another in a "World Series." Divisional play would be a mainstay in Major League Baseball up to the present day.
How "Alternate" Is It?
Unlike many alternate histories, the one presented here stays very true to our timeline. More realistically, an alternate timeline diverging in 1892 would likely, via the butterfly effect, generate a vastly different history thenceforth than our own. Though the close similarities in the timelines may not be very realistic, I needed to work from the actual team vs. team records in order to calculate season standings in the alternate timeline. Thus, the parallels were unavoidable in order for the calculations to remain accurate. Franchise relocations and expansions occur as in OTL (at least up until 1998).
Despite the fundamental similarities in the two timelines, simulated season records can vary substantially from the actual records, due to the presence or absence of strong or weak teams in a particular division. The largest increase in winning percentage was that of the Toronto Blue Jays in 2003, who in OTL finished 86-76 (0.531) but in the ATL finished 100-62 (0.617). The largest decrease in winning percentage was that of the Colorado Rockies in 2004, who in OTL finished 68-94 (0.420) but in the ATL finished 51-111 (0.315).
The Unbalanced Schedule
For each season, the total number of scheduled games per team is the same as in OTL, though the unbalanced schedule reapportions the games in favor of more intra-divisional play. The schedules used for each simulated season are as follows:
NL 1892, 1898-99: 12 teams, 154 games (20 games x 5 opponents in own division + 9 games x 6 opponents in other division)
NL 1893-97: 12 teams, 132 games (18 games x 5 opponents in own division + 7 games x 6 opponents in other division)
NL 1900-03, 1919; AL 1901-03, 1919: 8 teams, 140 games (28 games x 3 opponents in own division + 14 games x 4 opponents in other division)
NL 1904-18, 1920-61; AL 1904-18, 1920-60: 8 teams, 154 games (30 games x 3 opponents in own division + 16 games x 4 opponents in other division)
NL 1962-68; AL 1961-68: 10 teams, 162 games (23 games x 4 opponents in own division + 14 games x 5 opponents in other division)
NL 1969-92; AL 1969-76: 12 teams, 162 games (18 games x 5 opponents in own division + 12 games x 6 opponents in other division)
NL 1993-97; AL 1977-97: 14 teams, 162 games (13 games x 6 opponents in own division + 12 games x 7 opponents in other division)
NL 1998-present; AL 1998-present: 16 teams, 162 games (22 games x 3 opponents in own division + 8 games x 12 opponents in other 3 divisions)
Even if a team was in OTL unable to complete its full schedule (due to player strikes, rainouts, etc.), the full season has been simulated to show how the team would have finished had they played all their scheduled games. In many seasons, the altered schedule generates an entirely different postseason outcome, while in some others, the outcomes end up the same. Because the divisional setup and schedule for the AL from 1969-76 and 1979-93 are identical to those of OTL, the alternate season standings are almost the same as the true standings, differing only in that missed games have been filled in. However, ALCS results may differ, as they are based on average head-to-head winning percentage (this statistic will be explained further on) and not on the actual results. Though the basic NL schedule from 1969-93 was the same in OTL as in the ATL (18x5 + 12x6 for 1969-92 and 13x6 + 12x7 for 1993), the season standings differed more than the AL because the ATL places Atlanta and Cincinnati in the East and Chicago and St. Louis in the West.
Calculating Alternate Season Standings
The method by which I calculated the alternate season standings is derived almost directly from Don Mankowski on his website here. Although he simulated just 1941-45, it is his site and his technique which inspired me to simulate every season from 1892 to the present. The actual head-to-head winning percentage between any two teams in any one season is extrapolated onto the modified head-to-head schedule between the teams for that season. This is best described by Mr. Mankowski himself on his website (he is using the unbalanced 8-team, 154-game schedule):
If two teams split 22 games 11-11, we assume that they would be 15-15 against each other as divisional rivals, or 8-8 against each other if in opposite divisions. But, if a team went 18-4 (.818) against another, we'd project them to win 81.8% of 30 games and thus go 24.5-and-5.5. Since that looks silly, we'd round off and say 24-6 or 25-5. Similarly, we'd expect them to win 81.8% of 16 games, which works out to 13-3 more or less.
If the simulated head-to-head record amounted to exactly 24.5-5.5, rather than arbitrarily rounding to 24-6 or 25-5, I chose to always round the larger number up. If, for example, in 13 head-to-head games, the simulated record were 6.5-6.5, I rounded up the wins for the team which had a better actual head-to-head record that season. If this were also tied, I next consulted actual head-to-head runs scored that season, and then by the simulated average head-to-head winning percentage against all teams in the league (to be explained shortly). Tiebreaker playoff games were excluded from the actual season records and runs scored before I calculated the alternate season records. This is because these games were only added to the actual regular season because two teams completed their full schedule tied for a playoff spot, which would not necessarily have occurred in the alternate timeline.
Determining Postseason Winners and Breaking Ties
I used the following criteria to decide the winners of postseason series and tiebreaker playoff games, as well as playoff seed when two division winners had the same record and which way to round when simulated head-to-head records were tied but not whole numbers (e.g., 6.5-6.5). If the first criterion listed is not tied, there is no need to consider further criteria. If it is tied, I next looked at the second criterion, and if it is also tied, I went to the third, and so on.
1) Simulated head-to-head win-loss record
2) Actual head-to-head win-loss record
3) Actual head-to-head runs scored
4) Simulated average head-to-head winning percentage against league opponents
5) Actual average head-to-head winning percentage against league opponents
6) Simulated season winning percentage
7) Actual season winning percentage
8) Actual season runs scored-runs allowed ratio
In order to determine a World Series winner, I used a different strategy than for league division and championship series. Since there are no games played during the season between the two World Series teams (with the exception of interleague games, which are discussed later), head-to-head records and runs scored cannot be used as criteria to determine the winner. Instead, each team's overall performance against the other teams in its own league is used to predict a winner. Overall performance could be measured by a team's season winning percentage, but this may obscure the fact that the team happens to be in a particularly weak or strong division.
To better measure a team's overall season performance, the team's winning percentages versus every team in its league that year are averaged to produce the cleverly named "average head-to-head winning percentage." This statistic levels the playing field, so to speak, by effectively balancing an unbalanced schedule. For example, in an 8-team 154-game season, the weight of a 30-game record against a division rival is reduced and the weight of 16-game record against a non-division rival is increased so that they are weighted the same. The final result is a winning percentage similar to that of the team in OTL, if the team played a balanced schedule in OTL (in this example, 22 games versus each other team). Admittedly, even the average head-to-head winning percentage may obscure whether one league is on the whole stronger or weaker than the other, but, finding no viable way to accurately compare the leagues' strength, I deemed the average head-to-head winning percentage the most logical criterion to determine World Series winners for my alternate timeline.
Interleague Play (or lack thereof)
Interleague games are excluded from all extrapolations primarily because interleague play was never instituted in the alternate timeline. Since interleague play horribly complicated scheduling, I figured it was just as well that I didn't have to deal with it in calculating alternate records. Actual head-to-head interleague records are not considered when deciding alternate World Series winners, even if the two teams did play each other that season. With the exception of so-called "geographic rivals," a team plays only 3 games in a season against any particular interleague opponent. The results of so few head-to-head games make poor predictors for postseason series (or poorer than the typically more numerous intra-league head-to-head games, anyway).
Expansion and Realignment in the 1990s
The alternate timeline diverges most dramatically from our timeline in the 1990s. Instead of adopting a 3-division-per-league arrangement in 1994 in which the divisions did not all have an equal number of teams, the East-West format remained in place through 1997. (Presumably, the retention of the 2-division format and the lack of interleague play meant that Bud Selig failed to become acting commissioner in 1992 in the ATL. All the better for baseball, I say.)
In the alternate version of the 1998 season, four expansion teams (2 NL and 2 AL) were added to bring the total number of major league teams to 32. The two leagues were also realigned that season into four 4-team divisions each. In this manner, the awkward, ill-conceived arrangement of 16 NL and 14 AL teams in OTL was avoided. Since in reality the addition of 2 expansion teams in 1998 spurred talks of contraction only a few years later, it is more likely that the 4-team expansion would have occurred later than 1998 in the alternate timeline. However, I needed the actual data from Arizona and Tampa Bay for those years to properly determine simulated season standings, so a premature expansion was another unavoidable consequence of my strategy.
Thus, in the alternate 1998, the expansion Arizona Diamondbacks and Tampa Bay Devil Rays were added to the NL, and the expansion Portland Timbers and Washington Senators were added to the AL. The absence of Bud Selig as commissioner in the ATL allowed the Milwaukee Brewers to remain in the AL, their league of origin. Tampa Bay was instead placed in the same league as Arizona (as should have occurred in OTL in lieu of shifting Milwaukee to the NL). The presence of an expansion team in Washington prevented Montreal from relocating there in 2005, so the Expos stay put in my ATL. I use the actual stats of the Nationals from 2005 on to simulate how Montreal would have fared.
This departure from OTL complicated the calculations for the alternate standings from 1998 on. It's difficult to explain exactly how I derived the simulated records for Tampa Bay and Milwaukee, but essentially I estimated how these teams would have fared playing in the opposite league based on their season records in OTL. To simulate the records of the new Portland and Washington expansion teams, I averaged the actual records of the 14 existing expansion teams in OTL for each season of their existence and used this average winning percentage to estimate how well Portland and Washington would have fared against the other AL teams. For a more detailed description of the intricate calculations behind the season standings and postseason results, feel free to contact me, and I'll give you the breakdown.
Get on with it already
Thus, after intense (and at times aggravating) number-crunching, I've fleshed out an alternate history of Major League Baseball from 1892 to the present, complete with season standings and postseason series results. Due credit must be given to Baseball Reference for providing the essential data I used in my calculations. I am most grateful for the hard work that has been done by Sean Forman and his colleagues to compile that immense database. So, without further ado, here's what you all came here to see.