700 CLARK — RESEARCH

Starts have been shrinking for a century. One number refused to move.

After our last piece on the third time through the order, one question kept coming back. Cardinals starters rarely finish the sixth inning these days. How unusual is this? We went to the record to check — every major-league start from 1910 through 2025, rebuilt play by play.

Everyone says pitchers get pulled earlier than ever. By the third-trip measure, they are pulled about as often as they were 60 years ago. From 1950 through 1969, 84% of starts reached a third trip. Over the past decade, 84%. (Finishing the trip is another matter — the fourth-lap numbers below cover it.) What changed is why the one start in six ends early. Then, an early exit meant the pitcher was getting hit. Now it is usually the plan.

The full century says the same. Starts have collapsed by every measure of length: innings, batters faced, full trips through the order. Yet the share of starters who face a lineup a third time has barely moved. It was 88% in 1910. It was 85% in 2025. What has happened is that the hook has marched backward one lap at a time. The fifth trip through the order is gone. The last one belongs to David Price, who faced 37 batters for Detroit on September 6, 2014. The fourth trip is nearly gone. The third trip is where the retreat stands now. Starters still meet a lineup a third time. They just no longer finish the lap.

In 1910, three starts in four included a fourth trip through the order (1,850 of 2,456). One in four included a fifth. Pitchers who started games finished them, however long they ran. Dizzy Dean's 1934 season shows the old normal: 33 starts, 30 fourth trips, 12 fifth trips. In 2025 the entire league, across 4,859 starts, took 238 fourth trips and no fifth trips. Dean took more fifth trips by himself in 1934 than all thirty teams have taken since 2015.

A note on those old totals. A fifth trip usually meant a long game or extra innings, and the starter simply stayed for all of it. That was the era's custom. The man who started the game was expected to end it, however long it took.

The leaders tell the same story in one line. Robin Roberts took 38 fourth trips in 1953 (41 starts). Wilbur Wood took 38 in 1972 (49 starts). David Price led baseball with 28 in 2014 (34 starts). In 2024 the leader, Max Fried, took eight (29 starts). In 2025 it was Framber Valdez, with ten (31 starts). Even the workhorses have given up the fourth lap.

Through all of that, the top line barely bent. A third trip begins with the 19th batter of a game. The typical 2025 start ran 23 batters, four past the start of a third trip.

The trends match what fans believe about how the game has changed. The typical start lost 10 batters and more than two innings across the century. But the third-lap line moved on its own schedule.

The 1950s dip came from wild swings in how long outings ran. The same 1953 league that let Robin Roberts finish nearly everything also knocked one starter in six out before the 19th batter. Managers were not planning short starts; hitters were forcing them.

The peak came half a century later, and it is recent enough to feel strange. In 2011 and 2014, 95 of every 100 starts reached a third trip. Pitch counts had made outings uniform. Nearly every healthy starter got his six innings, and few got more. Chris Carpenter's Cy Young season in 2005 drew the era's ceiling: 23 fourth trips (33 starts), tied for second in baseball, and not one fifth. Adam Wainwright's 2014 fits the same mold: 19 fourth trips (32 starts), tied for fourth in all of baseball. Robin Roberts alone had doubled that figure in 1953.

The fall is the newest part. From 2015 to 2019, the third-lap share dropped from 92% to 84%. This time hitters were not forcing it. Teams began ending starts mid-lap on purpose. The two lowest shares ever came right after: 70% in the 60-game 2020 season, then 78% in 2021. The number has since settled near 85%. The result is that every modern third-pass number comes only from starts managers chose to extend.

A short start also changed its meaning. In the 1950s, hitters' OPS on the second pass in starts that ended early ran about .850 higher than in starts that continued. An early exit meant the pitcher was getting hit. In 2025 the difference was .212. More and more, teams plan the short start before the game begins.

Baseball even ran the experiment for us. From 1973 through 2021, the American League had a designated hitter and the National League did not. NL managers had a second, non-pitching reason to end a start: pinch-hitting for the pitcher. The two leagues reached the third trip at nearly the same rate in every era. The exit gap is where they parted. Before 1973, the NL's gap ran about 80 points higher than the AL's. The DH years flipped that: an AL early exit stayed a sign of trouble, while the NL's told less. Through the split years, the NL's early-exit gap ran 60 to 90 points of OPS lower. When the NL adopted the DH in 2022, the era averages closed to about 25 points apart. That is near even, though single seasons still swing more than that. A rule change moved this line on schedule, twice. That is strong evidence the line tracks why starts end, even though the two leagues were never perfectly level to begin with.

You can see the same change on the 2025 Cardinals: 8 fourth trips all season, four of them by Sonny Gray (32 starts, in his final St. Louis season). And Matthew Liberatore — the pitcher our last piece centered on — reached a third trip in 26 of his 29 starts in 2025. He reached a fourth in none. His third-pass record exists entirely inside the modern leash.

Our last piece measured the difference to be about 30 points of OPS in 2026 — the same starts, third trip against second. That number has a built-in problem, and we said so at the time. It comes only from starts a manager allowed to continue. Hook earlier, and the surviving third trips come from the better days. The measured difference could shrink without a single pitcher changing. Testing whether the hook bends the difference took a century of data.

The century's eras seem to answer it. From 1950 to 1969, managers hooked earliest: 84% of starts reached a third trip, and the measured difference averaged 96 points — the century's largest. From 2005 to 2014, managers were at their most patient: 94% reached, and the difference averaged 59 points — among the century's smallest. That pairing runs opposite the worry. The difference is measured only against second passes from starts that survived the hook. In the 1950s, surviving meant not getting hit — that is what chart 3 shows. The stretch was worth about 63 points in 1950–1969 and about 25 points in 2005–2014. Set it aside, and the two eras land at 32 and 35 points. The era contrast is really a portrait of how managers hooked, not of how pitchers held up.

Across 116 seasons, and across 840 team-seasons since 1998, there is no link between how many starts reached a third trip and what the third trip cost the pitcher.

So, back to the question we started with: why care about the third time through when starters barely finish the sixth? Because the third trip is the only deep lap left, and the OPS difference is the reason. When you see a third-pass number, read it as a pair: the difference, and how many starts managers let get there. In 2026 so far, that pair is 30 points at 84%. The first number passed every test we could throw at it. The second is the one a hundred years of baseball kept quietly changing.

Retrosheet play-by-play for every regular-season game, 1910 through 2025, using only games with complete accounts. Coverage is the record's, not ours: 69% of games in the worst year (1944), 100% every season since 1969. Definitions match our earlier pieces — the starter, each batter's pass count against him, and OPS built from the raw events. "Reached a lap" means the starter faced at least one batter that many times. In seasons without a designated hitter, the batters a starter faces include the opposing pitcher; that lowers every cell's OPS. It nearly cancels in the league-gap comparison. It cancels less well in the third-minus-second difference, because pitcher plate appearances fall more heavily on the second pass than on the third; that padding fades as the DH spreads. League splits classify each start by the pitching team's league. Interleague games — roughly one start in ten from 1997 through 2021 — follow the home park's DH rule, which blurs the league split slightly. The stretch figures in the closing section come from the same tables, split by whether the start survived to a third pass. As a check, the 2025 season computed this way matches our published 2025 baseline exactly. The charts draw from the same tables the sentences quote.