Baseball :: The Death of the Complete Game - Seven Decades of Pitcher Workload Data

Hello everyone!

There was a time when a starting pitcher who left before the ninth inning had some explaining to do. Finishing what you started was the job, not an achievement. Today a complete game is rare enough that it makes highlight reels, and the numbers behind that shift are some of the most dramatic in all of baseball's statistical record.

So let's look at what actually happened. Not the nostalgia, not the "pitchers were tougher back then" arguments, but the seven decades of workload data that show exactly how the complete game went from routine to nearly extinct.

1. What a Complete Game Actually Is

A complete game is credited when one pitcher throws every inning his team plays in the field. He does not have to win, and he does not have to pitch a shutout. He simply has to be the only pitcher his manager uses. In a home team's win it can be eight innings; in extras it can be twelve.

That definition has never changed. What changed is everything around it: how hard pitchers throw, how many arms a roster carries, how teams think about the third time a hitter sees a starter, and how much money is tied up in elbows. The statistic stayed constant while the sport moved out from under it.

2. Seven Decades of Disappearance

Here is the core of the story. The table below tracks league-wide complete games, what share of all team-games ended with one pitcher finishing, the average length of a start, and how often managers went to the bullpen.

Season	Complete games	Share of team-games	Innings per start	Relief outings per team
1970	852	21.9%	6.4	269
1980	856	20.3%	6.3	253
1990	429	10.2%	6.1	326
2000	234	4.8%	5.9	412
2010	165	3.4%	6.0	464
2015	104	2.1%	5.8	504
2025	29	0.6%	5.2	534
2026 (partial)	20	0.4%	5.1	521

Source: Baseball-Reference, Major League year-by-year pitching averages and season starter/reliever splits. 2026 figures are through roughly 159 games per team.

Read the middle column first, because it is the one that controls for league size. In 1970, better than one game in five ended with the starter still on the mound. By 1990 it was one in ten. By 2015 it was one in fifty. In 2025 it was roughly one in one hundred and seventy, and the 2026 season has been running lower still.

Plotted as complete games per 1,000 team-games, with 1950 included for scale, the collapse is easier to feel than to describe:

1950		403
1970		219
1990		102
2000		48
2010		34
2025		6
2026		4

Source: Baseball-Reference league totals; rate calculated as complete games divided by total team-games, times 1,000.

The 1950 bar and the 2026 bar are the same statistic measured the same way. One era finished roughly 40 percent of its games with a single arm. The current one finishes about 0.4 percent. That is a hundredfold change in a rule-stable category, which is close to unheard of in baseball's record book.

3. Why Starters Come Out Earlier

Four forces did most of the work, and none of them is about toughness.

The third time through the order. Hitters improve measurably each time they face the same pitcher in a game. Once teams could quantify that penalty, leaving a starter in for a fourth look at the top of the lineup stopped being brave and started being a measurable mistake.

Effort per pitch. Modern starters throw harder and spin the ball more aggressively than their predecessors, which buys strikeouts but spends stamina faster. A 1975 starter pacing himself through 130 pitches and a 2026 starter emptying the tank for 90 are not doing the same job.

Better bullpens. When the seventh-inning reliever is a legitimate power arm rather than a failed starter, handing him the ball is an upgrade, not a concession. That single change rewrote the manager's math.

Roster economics. Teams carry far more pitchers than they used to, and they use them. The chart below shows the average number of distinct pitchers each club sent to the mound in a season:

1970		15.1
1980		14.6
1990		18.6
2000		20.2
2010		21.2
2019		27.7
2026		28.8

Source: Baseball-Reference, pitchers used per season divided by number of clubs (24 teams in 1970, 26 in 1980 and 1990, 30 from 2000 on).

A 1980 club got through a season with roughly fifteen pitchers. A 2026 club needs nearly twice that. Injuries account for part of it, but the larger driver is design: the modern pitching staff is a rotating inventory, not a fixed unit, and no individual within it is expected to carry nine innings.

4. What Replaced It

The innings did not vanish, they moved. The interesting question is what the relievers who absorbed them are actually being asked to do.

Season	Relief outings per team	Outs per outing	Multi-inning outings	Saves per team
1970	269	4.6	148	37
1990	326	4.3	168	43
2010	464	3.1	113	40
2019	552	3.3	151	39
2025	534	3.3	148	40

Source: Baseball-Reference, Major League relief pitching splits by season, expressed per club.

Notice that relief outings roughly doubled while outs per outing fell from 4.6 to about 3.3. The reliever's job did not just become more common, it became narrower. A 1970 reliever was often asked for two innings; a modern one is frequently asked for three or four batters and nothing more.

The one number that has barely moved is saves, hovering around forty per club across all six decades. That is not a coincidence. Save opportunities are set by how often games are close in the ninth, which the sport's competitive balance keeps roughly constant no matter how the innings before it are divided.

5. Is It Ever Coming Back?

Probably not in any meaningful way, and the reason is that every incentive points the other direction. Nothing in the current rulebook rewards a manager for leaving a starter in, and plenty punishes him for it. Even the pitch clock, which shortened games considerably, did nothing to lengthen starts.

What survives is the record book, and it is a strange place to visit. Cy Young finished 749 games in his career. No active pitcher will approach a tenth of that. The entire thirty-team league now produces fewer complete games in a season than a single durable 1970s workhorse managed on his own, which makes the category less a statistic than a fossil.

That is worth knowing the next time you watch a starter get pulled at 95 pitches with a shutout going. It is not weakness and it is not bad managing. It is a hundred-year trend that finished arriving.

Final Thoughts

The complete game is the clearest example in baseball of a statistic outliving the conditions that produced it. Nobody banned it and nobody voted on it; teams simply discovered, one decision at a time, that the ninth inning belonged to somebody else. When a pitcher does finish one now, the crowd reaction tells you everything about how unusual it has become.

If you get to see one this season, appreciate it properly. At current rates, roughly one game in every two hundred and fifty will give you the chance.

#baseball #MLB #completegame #pitching #startingpitcher #reliefpitcher #bullpen #baseballstats #sabermetrics #pitcherworkload #inningspitched #baseballhistory #CyYoung #saves #pitchcount #baseballanalytics #MLBstats #thirdtimethroughtheorder #pitchingstaff #baseballdata