The Analytics Era: From Moneyball to the Modern Front Office
How the 2002 Oakland Athletics became the symbol of baseball analytics, where the ideas came from, and how they reshaped front offices, pitching and the home run, strikeout and walk game by 2019.
In 2002 the Oakland Athletics won 103 games and the American League West, finishing ahead of an Angels team that won 99 and a Mariners team that won 93, with one of the smaller payrolls in the majors. A book by Michael Lewis, Moneyball, appeared the following year and turned that season into a parable about how to win without money. The parable has been argued over ever since, but its central idea, that a club should hunt for value where other clubs are not looking, became the working assumption of nearly every front office in the game.
This article separates the story from the ideas. It looks at what the 2002 Athletics actually did, using the historical record for the numbers, where the analytical approach came from before Oakland, and how it changed the shape of play between 2002 and 2019 and after. It ends with the limits, because a good history of analytics should say what analytics did not and could not do.
What the 2002 Athletics actually looked like
The popular version of the Oakland story is that the club ignored batting average and bought on-base percentage. The record is more complicated and more interesting. The A's hit .261 as a team in 2002, just under the league average of .264. Their on-base percentage of .339 was fifth among the 14 American League clubs, behind the Yankees, Mariners, Red Sox and Angels. Where they stood out was in two other columns: they drew 609 walks, third behind the Yankees' 640 and the Mariners' 629, and they allowed 654 runs, the second-fewest in the league behind the Angels' 644.
They also stole only 46 bases, the fewest in the league. That is the part that matches the book's thesis. A stolen base is a gamble that costs an out when it fails, and Oakland's front office judged that the gamble rarely paid. Meanwhile their rotation was young and good: Barry Zito went 23-5 with a 2.75 earned run average and won the Cy Young Award, Mark Mulder went 19-7 and Tim Hudson went 15-9 with a 2.98 earned run average. Miguel Tejada, the shortstop, was the league's Most Valuable Player, and third baseman Eric Chavez hit 34 home runs with 109 runs batted in.
So the 2002 team was a good pitching staff, a patient lineup and a defense that kept runs off the board. It was not a lineup of strange bargains who somehow beat the odds. It did lose its best-known free agents after 2001: Jason Giambi, a former MVP who hit .342 in 2001, signed with the Yankees, and Johnny Damon signed with the Red Sox. Replacing them cheaply, and still winning 103 games, is what gave the season its reputation.
That summer also produced a streak of 20 straight wins, and the twentieth, against Kansas City, is one of the best-known games of the era: Oakland blew an 11-0 lead and then won on a walk-off home run by Scott Hatteberg, a former catcher the A's had made into a first baseman because he got on base. The streak was famous because it came from a club that had been talked about as an underdog. The season ended on a disappointing note: Oakland lost the division series to Minnesota, 2-3.
The idea before Oakland
The thinking behind Moneyball is older than Billy Beane's time in the front office. Branch Rickey, the executive who built the Dodgers of the 1940s, worked with a statistician, Allan Roth, and argued in print in the 1950s that on-base percentage and other overlooked measures said more about a team's offense than batting average alone. That argument sat in the margins of baseball for decades.
Two figures carried it forward. The first was Bill James, a writer who began self-publishing his Baseball Abstract in 1977 and who asked, over a long series of annual volumes, which of baseball's customary statistics actually predicted winning. The second was the Society for American Baseball Research, founded in 1971, which gave the people doing this work a home and whose initials James used to coin the word sabermetrics. Other writers built on both. The 1984 book The Hidden Game of Baseball, by John Thorn and Pete Palmer, assigned run values to every event on the field. Voros McCracken's 2001 research argued that pitchers have much less control over whether a ball in play becomes a hit than their earned run averages suggest, a finding that eventually changed how teams judge pitchers.
By 2002 the ideas were available to anyone who read the books and, increasingly, the websites. What made Oakland unusual was that its executives, including general manager Billy Beane and his assistant Paul DePodesta, used them to make decisions under budget pressure. A club with limited money has to buy something, and Oakland bought the skills it believed the market was undervaluing: patience at the plate, college hitters in the draft, and relievers whose style scouts disliked. Lewis's book turned that choice into a story.
The Red Sox hired Bill James as an adviser in the 2002-03 offseason, and the club won the 2004 World Series, ending a title drought that dated to 1918. It is easy to over-read that coincidence, but it shows the idea moving from the margins to the center of a large-market team. By the end of the 2000s, analysts had a seat at the table in most front offices.
The numbers behind a changing game
Between 1980 and 2025 the way runs are scored changed enough to show up in plain league totals. The table uses historical team totals for the American and National Leagues in each season. Plate appearances are approximated as at-bats plus walks, hit by pitch and sacrifice flies, so sacrifice bunts are excluded.
| Season | Runs per team-game | HR per team-game | Strikeouts per team-game | Walk rate | Strikeout rate | Complete games per 100 starts |
|---|---|---|---|---|---|---|
| 1980 | 4.29 | 0.73 | 4.80 | 8.3% | 12.7% | 20.3 |
| 1990 | 4.26 | 0.79 | 5.67 | 8.7% | 15.0% | 10.2 |
| 2002 | 4.62 | 1.04 | 6.47 | 8.8% | 17.0% | 4.4 |
| 2019 | 4.83 | 1.39 | 8.81 | 8.6% | 23.1% | 0.9 |
| 2025 | 4.45 | 1.16 | 8.36 | 8.4% | 22.3% | 0.6 |
The walk rate barely moves. Hitters did not suddenly start drawing more walks after Moneyball; the share of plate appearances that end in a walk was close to 8 or 9 percent in every year of the table. What moved was the home run and the strikeout. The home run rate went from 1.94 per hundred plate appearances in 1980 to 3.65 in 2019, and the strikeout rate went from 12.7 percent to 23.1 percent. In 2018, for the first time in major league history, the league recorded more strikeouts than hits, and that has been true every year since, including 2025.
Add the three true outcomes (home run, walk, strikeout) together and they accounted for 22.9 percent of plate appearances in 1980, 28.5 percent in 2002, 35.3 percent in 2019 and 33.8 percent in 2025. In those plate appearances nobody fields the ball, no runner moves up on a grounder and no outfielder runs anything down. The game became, on average, a little more predictable and a good deal less crowded with action, which is the complaint critics of analytics make most often.
The 2002 season sits midway along the curve. A year later the idea had a bestseller, and by 2019 the curve had run for another 17 years. The 2019 season had 1.39 home runs per team-game, the highest rate in the table, and 8.81 strikeouts per team-game. Three American League starters passed 250 strikeouts that year: Gerrit Cole (326), Justin Verlander (300) and Shane Bieber (259).
How the ideas changed the way the game is played
Several changes on the field trace back to the analytical approach, and some trace to other causes. It is worth separating the two.
Hitting for power. If a home run is worth much more than a single and an out is a cost either way, swinging for the fences becomes a sound trade even at the price of more strikeouts. This is the intellectual core of the three-true-outcomes game. The earlier surge in home runs, covered in the home run surge of the late 1990s and 2000s, had different causes, but it left the league used to a high-homer environment.
Pitching change. Starting pitchers throw fewer innings per start, and complete games, which were common in the 1970s (about 27 per 100 starts in 1973), have almost vanished: 0.9 per 100 in 2019 and 0.6 in 2025. Clubs found that the third time through a batting order, hitters fare better against a tiring starter, and they now bring relievers in earlier. In 2018 Tampa Bay began using a reliever as an opener to start games, an idea that came straight from that arithmetic.
Defensive positioning. The infield shift, which moves fielders to where a given batter tends to hit the ball, spread as teams learned from batted-ball data. Its spread was one of the visible faces of the analytics era, and it was eventually limited by a rule: in 2023 the league required two infielders on each side of second base, among the changes covered in the pitch-clock era article.
Player development. Teams use tracking systems such as Statcast, introduced in all parks in 2015, to measure exit velocity, launch angle, spin rate and sprint speed, and they build training programs around the data. Statcast is a measurement system, not a statistic; it tells a club how hard a ball was hit and at what angle, which a box score cannot.
Front-office structure. A club that once had a general manager, a scouting director and a few advisers now has analysts, data engineers, biomechanists and in some cases dozens of people who never see a game. The 2017 Astros, a rebuilt club that won the World Series, and the Rays, a pennant winner in 2008 and 2020 on one of the smaller payrolls, are the most-cited cases of analytics-driven organizations, and many other clubs followed the same path.
Rules respond: the 2023 reset
By 2019 the league was aware of its own numbers. Strikeouts outnumbered hits, the average game had less movement, and attendance, which was about 67.9 million in 2002, was about 68.5 million in 2019, essentially flat across 17 years with the same number of teams. The league's response came in 2023 with rule changes: a pitch clock, restrictions on defensive shifts, larger bases and limits on pickoff attempts. The stated aims were a faster pace and more action, and none of the changes banned a particular strategy outright.
The data shows an effect. League stolen bases per team-game rose from 0.47 in 2019 to 0.72 in 2023 and 0.71 in 2025, and the batting average rose slightly from .243 in 2022 to .248 in 2023. The strikeout rate came down from its 2019 peak, though it remains far above the levels of 1990. Check the league's own site and the rulebook for the details as they stand, since the rules continue to change.
The limits of the analytical approach
A fair account of analytics has to include what it did not solve. First, it cannot make a market permanently inefficient. The first clubs to use on-base percentage in the 1990s and 2000s gained an advantage, but once every club follows the same approach, prices adjust and the edge disappears. Oakland's own record shows it: the club made the playoffs often in the early 2000s, but after 2003 it won its division only four more times, in 2006, 2012, 2013 and 2020.
Second, short series reward luck. The 2002 Athletics lost a five-game series to a Twins team that won fewer games in the regular season, and the best regular-season record in a league does not guarantee a deep run. Analytics improves the odds of reaching October; it does not control what happens there. How the MLB Postseason Works explains why the format leaves room for upsets.
Third, the data are only as good as the questions asked of them. Counting statistics were once treated as more reliable than they were, and later metrics were sometimes treated the same way. A projection system gives a number; the front office still has to decide what to do with it. And the reaction against pure analytics is not only nostalgia: scouts still judge things that no sensor measures well, such as a young player's makeup and how he responds to failure.
Finally, the game itself is shaped by incentives. If a club learns that home runs and strikeouts are efficient, and every club learns the same lesson, the league's style changes for everyone. That is what the 2019 numbers show, and why the rule changes of 2023 were aimed at the style rather than at any single club.
A worked example: why on-base percentage matters
Suppose two hitters each have 600 plate appearances, with no hit by pitch and no sacrifice flies for simplicity. Hitter A has 540 at-bats, 60 walks and 162 hits, a .300 batting average and an on-base percentage of 222 divided by 600, or .370. Hitter B has 480 at-bats, 120 walks and 144 hits, a .300 batting average as well, and an on-base percentage of 264 divided by 600, or .440.
Both hitters bat .300. Hitter B reaches base 42 more times in the same 600 plate appearances, and he makes 42 fewer outs to do it (336 against 378). At roughly three outs per inning, 42 extra baserunners can be worth several more runs over a season, and that is before counting that a batter who walks forces the pitcher to throw more pitches. This is the reason on-base percentage became a core measure: it counts every way a batter avoids making an out. The example is hypothetical, but it is how the numbers work for real hitters.
Common questions
Was Moneyball only about on-base percentage? No. It was about finding undervalued skills and avoiding overpriced ones, and on-base percentage was the most visible example. The Athletics also valued pitching and defense.
Did the A's win because of analytics? They won 103 games with a strong rotation, a patient lineup and a good defense, and analytics helped assemble that group cheaply. It is not possible to separate the effect of the method from the talent of the players.
Does every team use analytics now? Nearly every club has an analytics or research department. The difference between clubs lies in how well they use what they find.
Is the three-true-outcomes game bad for baseball? That is a matter of taste. The numbers show it reduced balls in play; the league responded with rule changes in 2023.
Where to go next
For a rebuilding club at the other end of the standings, the 2019 Tigers lost 114 games, and their longer story is in the Detroit Tigers franchise history. The rules that followed the 2019 numbers are in the pitch-clock article. For the numbers themselves, use the leaders pages and the stat calculator to work out on-base percentage and OPS for yourself, and see the stats and analytics topic page for more reading.