Did the American Association of 1882–91 Achieve Parity with the National League? Evidence from Interleague Exhibition Games

The American Association (AA) began operation in 1882 as a major league, challenging the National League’s (NL) hegemony in professional baseball that had existed since the NL’s founding six years earlier. The AA had a mostly successful ten-year run until it merged with the NL in 1892. An interesting question is whether, at any time during those ten years, the AA achieved playing-field parity with the NL.

This article examines AA and NL parity by analyzing the results of preseason and postseason interleague games, of which there were many. The focus is on the five-year “middle” period of the AA’s existence, 1885–89, when parity was most likely to have existed. The principal source for identifying games is the New York Clipper, a national weekly newspaper with complete coverage of major league baseball including scores of all individual games.1 Critically, preseason and postseason exhibition results are reported, in addition to regular-season championship games.

We first present the relevant historical background of the AA and major league baseball in general in the 1880s and early 1890s. The second section discusses the analytical approach, including the use of exhibition game results for that purpose. Next we present the results of our interleague AA-NL comparisons, first at the aggregate level, then for individual clubs in both the AA and NL. The last section presents a summary and conclusions. We find that the evidence is consistent with AA-NL parity for the five-year period 1885–89.

 

The Louisville Colonels played in the American Association for the entirety of its existence, from 1882 to 1891. (WIKIMEDIA COMMONS)

 

BACKGROUND

The AA’s ten-year life can be split into three periods: initial: 1882–84; middle: 1885–89; and final: 1890–91. The AA began with six teams in 1882, expanding to eight the next year, and then to twelve in 1884. For its remaining seven years, the AA was an eight club circuit. It was an immediate threat to the NL, as its six charter cities had an average population of 375,328, almost twice the NL’s average of 189,692 for its eight member cities.2 After its successful first season, in 1883 the AA joined with the NL in the so-called Tripartite Agreement, along with the minor Northwestern League. It outlined various areas of interleague cooperation primarily for the purpose of mutual player contract recognition, and it was the first such agreement. In effect, with the Tripartite Agreement the NL conceded the AA’s major league status, constituting the first such “official” recognition. In 1884 it was renamed the “National Agreement” as it came to include more than one minor league.

The AA’s initial period mainly involved “getting up to speed” as a start-up major league. In 1882 and 1883, the NL still dominated the market for top players.3 AA clubs were therefore staffed mainly with NL retreads and players from minor league and independent clubs. The result was a lack of parity as demonstrated by NL dominance in interleague play. For example, in 1882 the AA lost all 21 preseason exhibition games against NL clubs.4 As David Nemec put it in his definitive history of the AA: “[in 1882] Association clubs… were no match as yet for established League clubs.”5 In 1883, the AA fared only slightly better, winning only four of 31 preseason games against the NL, a .129 winning average.6

The third year of the initial period, 1884, was aptly called “baseball’s wildest season” by historian William J. Ryczek.7 It was filled with turmoil caused by the entry of a third major league: the one-and-done Union Association. Ryczek notes that there were a total of 34 teams in the three leagues at various times, including several mid-season club failures, replacements, and relocations. The AA was part of the chaos, expanding to 12 members in 1884, plus a mid-season replacement for a failed club.8 The circumstances were hardly propitious for parity comparisons.

The AA’s final period began in 1890 with the entry of another one-and-done major league: the Players League (PL). It was the creation of the players’ union, The Brotherhood of Professional Base Ball Players, that was founded a few years earlier. Many observers consider it to be the strongest of the three majors that season, followed by the NL with the AA third. Unlike the Union Association of 1884, the PL was able to attract many top players from both the NL and AA. As Matt Albertson noted: “the Players League certainly had the best players in the country signed to contracts.”9 For example, when the 1890 season began, PL rosters included 107 1889 major leaguers, the NL had 63, and the AA only 36.10

For the AA, the establishment of the PL was a disaster.11 First, on November 14, 1889, the NL poached two of the AA’s best teams in preparation for the upcoming “war” against the PL: Brooklyn and Cincinatti. Brooklyn had finished first in the 1889 AA, and Cincinnati fourth. They replaced the NL’s two weakest clubs, Indianapolis and Washington, that in 1889 were the bottom two teams in the league. This unilateral act by the NL threatened the league’s viability and signaled that henceforth it was to be “every man for himself,” thereby rendering the then-existing cooperative National Agreement a dead letter.

To add insult to injury, the next day AA member Kansas City decamped to the minor Western League. Last, Baltimore resigned on November 29, later joining the minor Atlantic Association for 1890.12 Now down to only four clubs of the previous eight, the AA was desperate to expand its membership. It recruited three minor league teams from Rochester, Syracuse, and Toledo for 1890. A new start-up club was also added in Brooklyn, replacing the AA’s 1889 Brooklyn entry. While this restored the league to eight members, the AA’s new Brooklyn club was forced to compete with the NL’s recently recruited Brooklyn team, the previous year’s AA champion, and the Player’s League Brooklyn club. The net result was a considerably weaker lineup that included two second division 1889 AA clubs, three minor league 1889 clubs, and one de novo startup. As Harold Seymour put it, after this realignment, “the association [AA] could hardly claim major-league status.”13

Nevertheless, the AA was able to soldier through the 1890 season, albeit with two mid-season failures. First, the new Brooklyn entry went under on August 25 and was immediately replaced by the same Baltimore club that had left the AA nine months earlier.14 Next, the Athletic club of Philadelphia went bankrupt in early September. It was reorganized with a rag-tag assortment of low-end players, which finished the season by losing all 21 remaining games.15 It was then expelled from the AA in November.16 The 1890 championship was won by the same Louisville club that in the previous season had finished last.

For 1891, the AA was revamped…again. Clubs were added from Boston, Cincinnati, and Washington that replaced the three small market minor league “fill-ins” from 1890, plus a new Philadelphia franchise that replaced the expelled club.17 The new Boston and Philadelphia clubs had been members of the PL. Again, there was a mid-season failure, with Cincinnati replaced by the minor league Milwaukee club late in the season.

The key circumstance of the 1891 season for our purposes was that the AA was outside the newly reconstituted National Agreement. As before, it was a cooperative arrangement organized by the NL for contract protection that now included the minor Western Association. The AA originally joined but withdrew in February 1891, choosing instead to be an “outlaw” because of disputes with the NL. The main contention involved the reservation and allocation of players that had been on AA or NL clubs in 1889 but who had jumped to the PL.18 As such, the AA was blacklisted from playing games with National Agreement participants, and so in 1891 it had no preseason or postseason exhibition games with the NL.

In 1892, the AA merged with the NL although, given its weakened position, it would be more accurate to say that it was absorbed.19 Four AA clubs joined the eight NL teams to form a new 12-club organization with the remaining AA members paid to go “quietly into the night.” The former AA clubs finished 1892 at the bottom of the standings, occupying positions nine through 12. Although the Baltimore Orioles later gained fame with a highly successful run for the remainder of its NL membership, the other three AA “refugees” were tail-enders until the NL’s contraction in 1900. As Nemec put it: “so dreadful did the other three Association teams remain throughout the 1890s that it became hard to argue that the [National] League had not been superior all along.”20 We’ll now turn to addressing the truth of that statement.

 

 

PLAYING FIELD PARITY

We focus on interleague playing-field parity, i.e., whether teams in the two leagues were, on average, evenly matched in interleague play. This can be measured directly by performance in interleague games where wins and losses can be observed. Related concepts are parity of player quality and parity of overall play, i.e., are players and teams overall simply better in one league than the other. But both of these concepts have implications for wins and losses: if one league has better players and/or teams then, virtually by definition, its teams will win more often in interleague play. Thus, a “bottom-line” measure is win-loss records: parity means winning averages near .500 for both leagues in interleague play.

Playing-field parity between the AA and NL is examined by reviewing the results of preseason and postseason exhibition games involving clubs from the two leagues. We focus on the middle five years of the AA’s ten year existence: 1885–89. As noted above, this is the only period when parity might have existed and can be effectively evaluated. There were well over 300 such games, including the post-season World’s Championship Series that took place in each year under study. This is certainly a large enough sample to address the parity question.

The ideal situation for this analysis would have been each team in both leagues annually playing an equal number of games against every team. But, of course, nothing close to this ever happened. Exhibitions were arranged to attract fans and minimize travel cost, which often meant games between “natural rivals.” Examples included the Athletic (AA) and Philadelphia (NL) clubs (58 games); the Browns (AA) and Maroons (NL) of St. Louis (23 games); the Brooklyns and Metropolitans (AA) and the New Yorks (NL) (18 and 12 games, respectively); and the Baltimores (AA) and Washingtons (NL) (16 games). In fact, only a few times in any year did a team play all members of the other league. For example, the six AA clubs that were members for at least four years during 1885–89 averaged 6.2 NL opponents per year rather than eight, and the six NL clubs meeting the same criterion averaged only 4.3 AA opponents per year. Nevertheless, the large number of games actually played allows a reasonable, albeit imperfect, comparison.

Exhibition games, of course, differ from regular season championship games in that they do not impact the championship standings. Thus, at least some regular players may be less motivated. Also, regulars are more likely to be benched to recuperate from injuries or to simply rest. And these games provide an opportunity to test new players and give backups some playing time.21 All of these factors suggest that there may be less talent actually displayed on the field than during the championship season. Thus, exhibitions may not reflect a team’s full competitive capability.

On the other hand, given the significant number of these games over an extended period of several years, fans were apparently continuing to buy tickets. That implies that they viewed the games as something worth paying to see. As Scott Simkus put it: “exhibition games…must have been lucrative, otherwise [clubs] wouldn’t have booked them. …clubs would have had to put on a good show. Losing, especially as the result of lackadaisical play, is not a good show” (emphasis in the original).22 In addition, unexpected losses to inferior teams could damage brand names and raise concerns about “hippodroming,” i.e., games thrown to benefit gamblers.

There is no reason to think these issues regarding exhibitions, positive or negative, would be generally more prevalent in one league than the other. In other words, with our sizable sample, the overall impact on our between-league comparisons should be “a wash.” In technical terms, they should generate no statistical bias that might systematically impact the league average results, although they might increase the variance.

While playing field parity during the AA’s middle years is an unresolved question, there is general agreement among historians that the NL was superior regarding leadership and overall league management throughout the AA’s existence. It was the shortcomings in these areas that most likely led to the above-noted 1890 debacle and the AA’s ultimate downfall.23 Nevertheless, during 1885–89, the AA lineup was relatively more stable. It had six clubs that were members for all five years, while the NL had only four. In total, the AA had 11 different clubs during the period, while the NL had 13 (see below).

AA vs. NL EXHIBITION GAME RESULTS

Table 1 summarizes the aggregate results of AA vs. NL preseason and postseason exhibitions annually from 1885 to 1889. The top panel shows the five-year totals and the other five panels show the year-by-year results. These data include the end-of-season World’s Championship Series (WCS) that occurred each year. The WCS was not an official event sanctioned by the two leagues, but instead was bilaterally arranged between the two league champion clubs after the conclusion of the season on an ad hoc basis, like exhibition games. While it was generally expected to occur, it was not mandated and there was no fixed format for the series, including dates, number of games, or their location. Thus, it did not have the status or legitimacy enjoyed by the MLB World Series of today.

 

 

Our full sample consists of 354 exhibition games during the five years in question, including 44 WCS games, an average of 70.8 per year. The annual number varied from 58 (1888) to 81 (1887). There was a roughly even split between preseason and postseason games during 1885–87, then the last two years saw a small relative increase in preseason games. In total, there were 22 more preseason games.

The AA’s overall record versus the NL during 1885–89 was 170 wins, 172 losses, and 12 ties, a winning average of .497. This suggests that the outcome of these games was effectively a coin flip, i.e., on average outcomes were unrelated or “random” with respect to league membership. The preseason and postseason outcomes were similar, with winning averages of .481 and .515, respectively. The AA’s aggregate record in the 44 WCS games was 18–25–1, for a .419 winning average.24 Subtracting these from the five-year totals leaves a 152–147–11 record for the AA in 310 “regular” exhibitions, a .508 average, again not much different than a coin flip. While the WCS outcomes hint at NL superiority, that disappears when including regular exhibitions.

We can conduct a formal hypothesis test using our full sample (n=354) with the null hypothesis being parity, i.e., a mean winning average of .500. We can use the normal (bell-shaped) approximation to the underlying binomial distribution.25 The expected number of wins is 177 (.5*354) and the actual number is 176 (counting ties as a half win and half loss). The resulting test z-score is -.00032, less than .1 percent of the critical z-score of -1.64 at a liberal 10% rejection level. This is strong support for parity.

Looking at the annual results, the AA had a winning averages above .500 in three of the five years: 1885 (.529), 1886 (.551), and 1888 (.552). These were offset mainly by a weak 1887 record of 32–47–2, a winning average of .405, resulting in an overall five-year average of just under .500. These data suggest that the AA might have been, on average, slightly better than the NL in those three years.

In the subsample consisting of the ten preseason and postseason periods, the AA was above .500 versus the NL in three, under .500 in four, and the other three were an even split. Individual AA averages ranged from a low of .392 in the 1887 preseason to a high of .676 in the 1886 postseason. We can also conduct a formal hypothesis test using this subsample, again with the null hypothesis being parity, i.e., a mean winning average of .500. The actual mean of the ten subsample won-loss averages is .497 and their standard deviation is .0876. With the sample size of ten, the resulting t-statistic is -.051.26 To reject the null hypothesis at a liberal 10 percent confidence level would require a t-statistic of at least -2.26. As our actual t-statistic is only about two percent of this value, we are not close to rejecting the null hypothesis of .500. Again, this provides strong support for parity.

In sum, the AA’s winning average for all preseason and postseason exhibition games during 1885–89 was slightly under .500. In three of the five years it was above .500 and in seven of the ten preseason and postseason periods the winning average was .500 or higher. Overall, the aggregate record is consistent with AA-NL parity on the playing field for these five years, unless one gives significant disproportionate weight to the WCS results.

We can also look at the record at a more disaggregated level, i.e., how each club in the AA and NL fared against the opposing league. This gets somewhat “messy” because of the unstable membership of both leagues. The AA and NL each had eight members in every year of the 1885–89 study period. The AA had six “mainstay” clubs that competed in all five seasons: the Athletic, Baltimore, Brooklyn, Cincinnati, Louisville, and St. Louis (Browns) clubs. The remaining two AA slots were occupied at various times by five clubs that were members for three or fewer years: Pittsburgh, Cleveland, Columbus, Kansas City, and Metropolitan (New York). These “part-timer” AA clubs were generally second division finishers, with a combined winning average of .405.

The NL had only four clubs that were members in all five years: the Bostons, Chicagos, New Yorks and Philadelphias. Detroit and Washington were members for four years. We’ll refer to these six clubs as the NL “mainstays.” The remaining slots were occupied at various times by seven clubs that were members for three or fewer years: Pittsburgh, Buffalo, Cleveland, Indianapolis, Kansas City, Providence, and St. Louis (Maroons) clubs. Like the AA part-timers, these were generally second division finishers, with a combined winning average of .395.

Accordingly, the AA had a total of 11 clubs during 1885–89 and the NL had 13. The Pittsburgh club moved from the AA to the NL in 1887; the Clevelands moved from the AA to the NL in 1889; and the Kansas Citys moved from the NL to the AA in 1888 after an intervening year in a minor league.27

Table 2 below shows the records of the 11 AA clubs versus the NL during 1885–89, sorted by years of membership, then by games played. The St. Louis and Athletic clubs played the most games, at 90 and 64 respectively, between them accounting for over 40 percent of interleague games. St. Louis was involved in the WCS from 1885 to 1888 and played several games against their intra-city NL rival Maroons in 1885 and 1886, while the Athletics played numerous games against their intra-city rival NL Philadelphias. Including these two, a total of five AA clubs had at least 30 interleague games during our study period. Only one of the six mainstay members had a winning average under .500 against the NL: the Athletics at .461. The other five-year members had averages ranging from .510 to .595. Combined, these six clubs had a winning average of .530 against the NL. In contrast, the five AA “part-timers” as a group had a winning average of only .325 against the NL, ranging from .250 to .429. All had averages lower than the Athletics’ .461. Thus, the aggregate AA winning average of slightly below .500 noted above can be attributed mainly to the poor performance of the part-time clubs, which also did poorly against the AA mainstay clubs.

 

Table 3 below shows the records of the 13 NL clubs versus the AA during 1885–89, again sorted by years of membership, then by games played. The Philadelphias and New Yorks played the most games, 72 and 48 respectively. The Philadelphias played numerous games against the Athletics, and the New Yorks played many games against their two local AA rivals: the Brooklyns and Metropolitans. Like the AA, five NL clubs had at least 30 interleague games. Of the six NL mainstay clubs that were members for at least four years, two had winning averages against the AA of over .600 (Boston and New York), both greater than any AA team winning average against the NL. On the downside, two clubs in this group had winning averages under .500: Washington at .480, and Chicago, whose .426 winning average is lower than any AA mainstay member versus the NL. Thus the variation in club winning averages vs. the other league was greater in the NL. Overall, these six mainstays have a .535 winning average vs. the AA, slightly larger than the .530 figure for the mainstay AA group vs. the NL. The seven “part timer” clubs in the bottom group of Table 3 with three or fewer years as members had an aggregate winning average of .431, with Cleveland exceeding .500, at .583.28 Note that the NL’s part-timer group average against the AA is more than 100 points greater than the .325 average of AA’s part-timer group against the NL. This is the major difference between the two leagues.

 

 

SUMMARY AND CONCLUSIONS

We examine the question of playing-field parity between the American Association and National League during 1885 and 1889, the “middle” years of the AA’s ten-year existence. This is when parity is most likely to have existed, given the AA’s history before and after this period. A total of 354 interleague exhibition games were played during these five years, including preseason and postseason games and the post-season World’s Championship Series games in each year.

In sum, the results are consistent with AA parity with the NL. The AA’s aggregate record against the NL was 170–172–12 for a winning average of .497, i.e., essentially breaking even. In three of the five years AA clubs had above .500 averages against the NL. Formal tests of the hypothesis of parity (winning average=.500) for the full sample and the subsample of ten pre/post seasons do not come close to rejecting the hypothesis. The six mainstay clubs of both leagues had above .500 winning averages against the other league that were about the same, the NL’s being slightly higher. The main difference between the two leagues was among their weaker “part-time” member clubs, those with three or fewer years of membership. In both leagues, this group had records significantly below .500 against the other league, with the AA’s part-timers performing below the NL’s part-timers in interleague play. In terms of parity, this was the main difference between the two leagues.

The AA’s on-field performance during 1885–89 held some promise for long term success. But it’s weak leadership and generally haphazard management relative to the NL rendered it unable to cope with the major threat posed by the 1890 Players League. As a result the 1890 and 1891 seasons were largely failures, both on the playing field and financially. Then, in 1892, the weakened AA was absorbed by the NL and its ten-year run came to an end.

WOODY ECKARD, PhD is a retired economics professor living in Evergreen CO with his wife Jacky and their two dogs Petey and Violet. Among his academic publications are 13 papers on sports economics, five of which relate to MLB. More recently he has published in SABR’s Baseball Research Journal, The National Pastime, and Nineteenth Century Notes. He is a Rockies fan and a SABR member for over 20 years.

 

Notes

1. The Clipper is available from the Digital Newspaper Collections at the University of Illinois Library: https://idnc.library.illinois.edu/?a=d&d=NYC18771006.2.13&e=——-en-20–1–txt-txIN———-.

2. Based on 1890 U.S. Census data. The AA cities were Allegheny City (Pittsburgh), Baltimore, Cincinnati, Louisville, Philadelphia, and St. Louis. The NL’s 1882 lineup was Boston, Buffalo, Chicago, Cleveland, Detroit, Providence, Troy, and Worcester. Even without its two smallest cities, Troy and Worcester, the NL’s average population was still substantially less at 233,750. The population for the Allegheny City club includes Pittsburgh. At that time they were independent polities separated only by a river, akin to Brooklyn and New York at that time.

3. For example, see Richard Hershberger, “The First Baseball War: The American Association and the National League.” SABR Baseball Research Journal 49, no. 2 (2020), 115–25.

4. Edward Achorn, The Summer of Beer and Whiskey: How Brewers, Barkeeps, Rowdies, Immigrants, and a Wild Pennant Fight Made Baseball America’s Game (New York: Public Affairs, Perseus Books Group, 2013), 29; and David Pietrusza, Major Leagues: The Formation, Sometimes Absorption and Mostly Inevitable Demise of 18 Professional Baseball Organizations, 1871 to Present (lemurpress.com: Lemur Press, 2020), 72.

5. David Nemec, The Beer and Whiskey League: The Illustrated History of the American Association—Baseball’s Renegade Major League (Guilford CT: The Lyons Press, 2004), 26.

6. Nemec, The Beer and Whiskey League, 46; and Achorn, The Summer of Beer and Whiskey, 50–51.

7. William J. Ryczek, Baseball’s Wildest Season: Three Leagues, Thirty-Four Teams and the Chaos of 1884 (Jefferson NC: McFarland & Company, 2023). See also Justin McKinney, Baseball’s Union Association: The Short, Strange Life of a 19th-Century Major League (Jefferson NC: McFarland & Company Inc., Publishers, 2022); and Richard Hershberger, “The Union Association War of 1884”, SABR Baseball Research Journal, Vol. 53, No. 1 (Spring 2024).

8. The National League was “above it all” with the same eight-club lineup as in 1883, and all successfully completed the season.

9. Matt Albertson, “Introduction and Context of the Players League Formation,” in Base Ball’s 19th Century “Winter” Meetings: 1857–1900, Jeremy K. Hodges and Bill Nowlin, eds. (SABR, 2018), 285.

10. John Bauer, “The Establishment Responds: Winter Meetings 1889–90 (NL/AA),” in Base Ball’s 19th Century “Winter” Meetings: 1857–1900, 277.

11. Rob Bauer, Outside the Lines of Gilded Age Baseball: The Origins of the 1890 Players League (robbauerbooks@gmail.com, 2018), 223–24.

12. Nemec, The Beer and Whisky League, 187. Toward the end of the 1890 season, Baltimore re-joined the AA, replacing the failed new Brooklyn entry.

13. Harold Seymour and Dorothy Seymour, Baseball: The Early Years (New York: Oxford University Press, 1994), 149.

14. Nemec, The Beer and Whisky League, 197.

15. John Bauer, “Three Divides into Two: The 1890 Winter Meetings,” in Base Ball’s 19th Century “Winter” Meetings: 1857–1900, 293. See also Clifford Blau, “The 1890 Athletic Club: The Worst Team in MLB History,” Seamheads.com, October 9, 2012, https://seamheads.com/2012/10/09/the-1890-athletics-the-worst-team-in-major-league-baseball-history/, accessed July 1, 2024.

16. John Bauer, “Three Divides into Two,” 293.

17. John Bauer, “Three Divides into Two,” 296.

18. John Bauer, “Three Divides into Two,” 297.

19. Seymour and Seymour, Baseball: The Early Years, 262.

20. Nemec, The Beer and Whisky League, 240.

21. These players, however, might be highly motivated, viewing the games as “tryouts.”

22. Scott Simkus, Outsider Baseball: The Weird World of Hardball on the Fringe, 1876–1950 (Chicago: Chicago Review Press, 2014), 22.

23. For example, see Nemec, The Beer and Whisky League, 242–44.

24. As the Browns were AA champions in four of the five years, the WCS results alone tell us little more than where the Browns stood against the NL champions, i.e., it’s an unrepresentative sample.

25. The binomial is a discrete probability distribution that applies to a situation with n “trials” (games); two possible outcomes at each trial: “success” (win, with probability p) or “failure” (loss, with probability q); and a constant success probability p. In our case, n=354 and p=q=.500 assuming parity. The expected value is n*p=177, with ties counting as a half win and half loss, and the standard deviation is (n*p*q)½=9.41. The sample size is well above the minimum threshold suggested for applying the normal approximation.

26. The calculation is t=(.497-.500)/(.0876/10½).

27. Upon moving to the NL, the Allegheny City winning average decreased from .584 to .444, and the Cleveland average increased from .379 to .462. For Kansas City, the intervening minor league year makes a comparison inappropriate.

28. Without the “outlier” Cleveland, the NL’s part-timer average versus the AA was .401.