Flu-Like Symptoms: The Biggest World Series Upsets

Image credit: USA Today Sports

The 1987 Twins are a famously not-great team to win the World Series. They ended a 63-year championship drought for the Washington/Minnesota franchise, but they didn’t have a great regular season. They won the AL West with an 85-77 record that would’ve been fifth in the AL East, trailing the 98-64 Tigers, 96-66 Blue Jays, 91-71 Brewers, and 89-73 Yankees.

Further, the Twins were outscored during the regular season. They scored 786 runs (eighth in the 14-team American League) while giving up 806 (sixth-most). The average American League team scored 4.90 runs per game in 1987; the Twins scored fewer (4.85) and gave up more (4.98). Yet they beat the Tigers handily in the ALCS, four games to one, and topped the Cardinals in a seven-game World Series.

That got me thinking. This year’s World Series has two fairly evenly matched teams. If the regular season had ended at the All-Star break, there would have been no question that the Dodgers and Astros were the class of their respective leagues. The 61-29 Dodgers led the NL by 7.5 games and the 60-29 Astros were 10 ahead of the Red Sox in the AL. They each cooled down after the break, but both won over 100 games. Either team winning wouldn’t be an upset.

But back to those 1987 Twins. Their World Series victory, like that of the Astros in this year’s ALCS, occurred with home teams winning every game. (This was before home-field advantage in the World Series was determined by the best regular-season record, or by the winner of the All-Star game. This was the scientifically rigorous “even or odd years” era.) How much of an upset was that?

Here at BP, we calculate each team’s wins based on Bill James’ Pythagorean Theorem of Baseball, which uses runs scored and allowed. We call this “first-order wins.” We also calculate “second-order wins,” which considers how many runs teams would be expected to score and allow, given their underlying statistics, and plugs those into the Pythagorean calculation. Our “third-order wins” adjusts the second-order wins for quality of opponent. You can see first-, second-, and third-order win percentages for each team this year by selecting Standings on the black ribbon at the top of this page.

The 1987 Twins’ first-order record was 79-83. That makes sense; they were outscored. Their second-order record was 77-85. And their third-order record was 76-86. So yeah, a not-great team.

But the 1987 Cardinals finished better than their runs scored and allowed would suggest, too. They were 95-67 but only 92-70 by first-order wins. Their second-order record was considerably worse: 82-80. Their third-order record was the same, 82-80. They finished second in the league with 798 runs scored, but their True Average—which measures total offensive value scaled to an average of .260—was .259, tied with the Reds for fourth in the NL, well behind the Mets (.273) and Giants (.264). And while they allowed the third-fewest runs in the league, their Deserved Run Average—which reflects all contributions of pitchers to the run-scoring that occurs around them—was 4.79, seventh in the league.

So the Cardinals-Twins series wasn’t the mismatch it appeared to be on paper, or at least if the paper contained only actual W-L and maybe Pythagorean. Looking at third-order wins, the 1987 World Series saw a team with 76.4 third-order wins in the regular season beat a team with 82.5. That’s an upset, to the tune of 6.1 wins, but that’s not a huge outlier. The 1987 World Series featured a not-great team beating a team whose record was a lot better than its underlying statistics.

What are the big upsets? Which World Series victories were the most improbable?

To answer this, I looked at the third-order regular-season wins of every World Series competitor, calculated the difference between the winner and loser, and sorted from the low to high. We have complete third-order records dating back to 1958. So here are the biggest World Series upsets since the year Pizza Hut was founded:

(“3-OrdW” is third-order wins. “Diff” is difference between the loser’s and the winner’s third-order wins. “Reg Seas Diff” is the difference between the loser’s and the winner’s actual wins during the regular season.)

Let’s break those down, with regular-season records in parentheses. Note that in many cases, the difference in third-order wins was pretty close to the difference in actual wins. But not always!

1969 Mets (100-62) def. Orioles (109-53), 4-1. New York won nine fewer games than the Baltimore during the regular season. The Orioles were a game worse than their first-order record, while the Mets were eight games better, and the gap widened to a whopping 26 games with underlying stats included. A year earlier, the Mets finished 73-89, which was the best record in franchise history. Everything about the Mets’ season, from winning 100 games to sweeping the Braves (one more third-order win) in the first NLCS to beating Baltimore, was an upset.

2011 Cardinals (90-72) def. Rangers (96-66), 4-3. A wild series, with three one-run games and questionable managerial moves galore. It pitted a Cardinals team that outperformed a little during the regular season against a Rangers team that underperformed a lot.

2006 Cardinals (83-78) def. Tigers (95-67), 4-1. The 1987 Twins, discussed above, won only 85 games. That’s two more than the 2006 Cardinals, who had a sub-.500 third-order record. The Tigers were a Wild Card team that year and outperformed their third-order wins by two. The division-leading Cardinals outperformed theirs by six.

1990 Reds (91-71) def. A’s (103-59), 4-0. Simply by regular-season record, the Reds were underdogs, winning 12 fewer games than the A’s. This, and the next three series listed, are cases where the raw winning percentage and the third-order calculations concur.

2003 Marlins (91-71) def. Yankees (101-61), 4-2. Yes, the Bartman Marlins. The Marlins punched above their weight in the regular season, largely due to a 30-23 record in one-run games that enabled them to outperform their 87 first-order wins.

1974 A’s (90-72) def. Dodgers (102-60), 4-1. More than anything else, this series is remembered for one play: Dodgers reliever Mike Marshall picking off A’s “designated runner” Herb Washington with one out in the ninth inning of the second game and Los Angeles clinging to a 3-2 lead. It’d be the Dodgers’ only win. Curt Gowdy and Vin Scully on the call:

1985 Royals (91-71) def. Cardinals (101-61), 4-3. Hey, Cardinals fans: Enough with the Denkinger whining. You’re on this list as winners three times.

2008 Phillies (92-70) def. Rays (97-65), 4-1. The Rays, who entered the season as 150-to-1 longshots to win the World Series, were favorites by actual record and even more so by third-order wins. There are bookmakers who took preseason action on the Rays and owe their continued employment to the 2008 Phillies. The Rays played pretty true to their underlying performance during the year, but the Phillies managed to be third in the NL in both runs scored and run prevention while being fifth in TAv and ninth in DRA.

1963 Dodgers (99-63) def. Yankees (104-57), 4-0. The second sweep on this top-10 list. The Dodgers outperformed their first-order record by seven games and their third-order wins by 10. The teams weren’t as closely matched as their regular-season records would suggest.

1982 Cardinals (92-70) def. Brewers (95-67), 4-3. Separated by only three games in the regular-season standings, the gap widens to nearly 10 as the Cardinals had 85 third-order wins. They allowed the fewest runs in the NL despite being eighth in FIP and sixth in DRA.

***

This year? The Dodgers had 105 third-order wins and the Astros had 102. There’s no upset brewing either way.