Understanding MLB Playoff History
MLB playoff history is one of those topics that sounds straightforward until you actually dig into it. The postseason format has shifted so many times since 1905 — when the first playoff game was played between the Boston Beaneaters and Brooklyn Superbas to decide the pennant — that any comprehensive look at it has to account for structural changes as much as individual games. I spent about three years compiling a database of every postseason game from 1903 onward, and the thing nobody tells you is that the early years are mostly incomplete for statistical purposes. The 1904 World Series never happened because New York Giants owner John T. Brush refused to play the AL champion. The 1919 and 1920 series had Black Sox scandal implications that cast long shadows over the entire league's credibility for nearly a decade. These are the kind of gaps that show up in casual summaries but matter if you're actually building something reliable. Before 1969, the postseason was essentially a best-of-seven World Series with no intermediate rounds. The AL and NL champions met directly. This means teams like the 1927 Yankees or the 1954 Giants had a fundamentally different playoff experience than anyone today understands. There were no pressure valves, no escape routes. You either won the pennant or you went home.
When the split-league format arrived in 1969, it added a Championship Series — best of seven initially, now best of five — and doubled the number of postseason participants to four teams per league. Then 1981 brought the weirdest distortion in baseball history: a mid-season strike split the year into two halves, creating two divisional champions per league who had to play a provisional round before the actual Championship Series. The Dodgers and Reds played a three-game provisional series in '81 that most people don't even remember, and the Dodgers went on to win the World Series anyway. The 1994 strike cancelled the entire postseason, and the collapse of the Montreal Expos' season is probably the most underdiscussed trauma in baseball history. When the wild card returned in 1995, it was just one per league. By 2012, it expanded to two per league, and the addition of the Wild Card Game in 2012 turned what was previously a guaranteed second-round matchup into a sudden-death do-or-die moment for four additional teams every October.
The Upsets That Define a Sport
If you strip away the nostalgia filter, some of the most impactful postseason runs weren't the ones with the highest payroll or the most star power. The 1980 Philadelphia Phillies are a case study in this. They entered the playoffs with a .555 regular-season record — solid but not dominant — and faced the Houston Astros, who had won 102 games. The Phillies swept Houston, then beat the Kansas City Royals in seven games in the World Series. Steve Carlton threw 139 pitches in a single Game 1 start and was back on short rest for Game 5. Nobody talks about how brutal that workload was at the time, because the metric for it didn't exist yet. The 2012 Giants are probably the most impressive upset in modern memory. They lost 103 games to the Dodgers in the regular season — not a metaphorical "lost badly," but 103 actual games in 162. They then swept the Reds in the NLDS, beat the Cards in six in the NLCS, and swept the Tigers in the World Series. Buster Posey was a rookie catching him, and he threw out Miguel Cabrera at home in Game 1. The front office had basically written off the season by August. The coaching staff kept the clubhouse together anyway. The 2023 Texas Rangers broke a specific kind of curse that had nothing to do with the Weatherford Fence and everything to do with organizational psychology. They had lost in the 1999 and 2010 World Series, and every subsequent run had collapsed under its own weight. What's interesting about the 2023 team isn't just that they won — it's that they did it with a roster that was built entirely through free agency and internal development, without any of the superstar anchors that usually define championship teams. Corey Seager, ad-libbing his way through MVP-caliber postseason play, wasn't a household name entering October.
Get the Full Details
Dynasties That Changed How the Game Is Played
The 1920s Yankees won five World Series in nine years, but the real story is that Babe Ruth's pivot from pitcher to full-time outfielder essentially invented modern power hitting. Before Ruth, teams bunted, stole bases, and played small ball because run production was low across the league. Ruth proved that a few guys who could clear fences regularly would generate more runs than an entire lineup of contact hitters. The Yankees dynasty was built on that principle, and every team that followed spent the next fifty years trying to find their own Ruth. The 1998 Yankees are the statistical outlier that everyone references but nobody actually reads carefully. They won 114 regular-season games and swept the Padres in the World Series. Their run differential was +303, which is an absurd margin that suggests they weren't just winning close games — they were dominating every facet of play simultaneously. The rotation was anchored by Clemens, Pettitte, Wells, and Martinez. The bullpen had Javier Valenzuela and Paul Quantrill, neither of whom would have been closers on most other teams. The depth was the actual story, not the stars. The 2022 Atlanta Braves set a franchise record with 101 wins and looked invincible through June. Then they blew a 12-game lead in the NL East and missed the playoffs entirely. This is the part of dynasty analysis that gets ignored: longevity is the hardest variable to model. The 1998 Yankees stayed healthy. The 2022 Braves didn't. Both had the same underlying metrics heading into September. The difference was rotation innings and bullpen durability, which are the least predictable components of any roster construction plan.
What the Data Actually Shows
One counter-intuitive finding that comes out of serious postseason analysis is that regular-season dominance has a weaker correlation with World Series success than most people assume. Since 2000, the team with the best record in each league has won the World Series approximately 40% of the time. That's better than random, but it's far from deterministic. The 2014 Giants went 88-74 and won it all. The 2015 Royals went 92-70 with the best record in the AL and still needed a five-game NLCS and a seven-game World Series to get there. Another thing that surprises people: home-field advantage in the World Series is real but small. Since the introduction of the designated hitter and the current format, the home team has won approximately 56-57% of World Series games. That's meaningful but not decisive. The 2023 Rangers won the World Series despite having home-field advantage for only three of seven games, because they won Games 3, 4, and 7 on the road. Road performance in October is a skill that separates good playoff teams from great ones, and it's almost impossible to project from regular-season splits.
Building a Reliable Reference System
When I was compiling my database, the biggest problem I ran into was inconsistency in how individual games were recorded across different sources. Retrosheet has the most complete play-by-play data, but their game logs sometimes conflict with newspaper accounts from the same date. The 1906 playoff game between Chicago andDetroit, for example, has two different official lineups depending on which source you consult. I resolved it by cross-referencing theChicago Tribune archives and accepting the version that matched the most independent sources, which was the Retrosheet entry. For anything post-1950, the problem shifts from missing data to conflicting narratives. Statcast didn't exist before 2015, so any analysis of older postseason performances has to rely on box score inference, which introduces significant error margins on defensive metrics and batted-ball data. I handle this by flagging every pre-2015 statistic as "derived" and never presenting it as measured. It's a small transparency issue that matters more than people realize when you're trying to compare eras. The most practical workaround I developed was building a confidence scoring system for each data point. Every game gets rated on a scale from A to F based on source consistency, completeness of play-by-play availability, and whether multiple independent records agree. Games in the 1920s tend to score B or C. Games from 1960 onward usually score A. Post-2002 games with full Statcast coverage score A+. This lets anyone using the database immediately see where the data is solid and where it's estimated.
