Aging Curves: When Hitters and Pitchers Peak and Decline

By 4Baseball Staff · November 5, 2026 · 10 min read

What baseball aging curves show about when hitters and pitchers peak and decline, why survivorship bias makes the raw averages lie, and how to use the curves without over-trusting them.

Ask when a baseball player peaks and you will hear a number: 27, say, or 28. That number comes from aging curves, graphs that follow players season by season and show how production rises, levels off and falls. The idea is simple. The execution is surprisingly easy to get wrong, because the most obvious way to draw the curve produces a graph that says players barely age at all.

This article builds a curve two ways from the Lahman database, hitters and pitchers from 1990 through 2025, and shows why the two ways disagree. It explains survivorship bias with real numbers, covers what researchers generally find about hitting skills and pitching skills, looks at the players who beat the curve, and ends with practical advice for fantasy owners and anyone reading a projection. The central lesson is that an aging curve describes a group, and a group does not tell you what one player will do.

What an aging curve measures

An aging curve plots a performance measure against age. The measure might be on-base plus slugging (OPS) for a hitter, or earned run average and strikeouts per nine innings for a pitcher. The age is usually the player's age during the season, and there is more than one way to count it: many analysts use age on a fixed date in the middle of the season, while a simpler count subtracts the birth year from the season year. This article uses the simpler count, which can differ from a player's real age by up to a year.

There are two basic ways to build the curve. The first averages every player's performance at each age. The second, usually called the delta method, looks at the same players in consecutive seasons and averages how much each one changed from one year to the next. The distinction matters more than any other choice in the whole subject, and the rest of this article is largely an explanation of why.

The easy method and what it shows

Start with the easy method. Take every hitter from 1990 through 2025 with at least 400 plate appearances in a season, compute his OPS, and average by age. The numbers come out like this.

Read that table at face value and you would conclude that hitters get slightly better every year until their mid-thirties and never decline. Nobody who has watched baseball believes that. The table is accurate as arithmetic and badly misleading as a description of aging, and the reason is that the groups at each row are not the same people.

Survivorship bias: the players who are still there

Players do not stay in the sample by luck. A hitter who declines at 31 stops getting 400 plate appearances at 33 and disappears from the 33-year-old row. The only 36-year-olds in the table are the ones who were still good enough to play every day. So the older rows are filled with survivors, who were better than average to start with, and the average of survivors stays high even while each individual is declining.

You can see this directly. Take only the 91 hitters in the data who still had a 400-plate-appearance season at age 38 or older, and look at their OPS at earlier ages.

The group that lasted was far better than average in its twenties, around .864 to .878 at the height of its career, and by 38 it had fallen to the same .791 as everyone else. That is a decline of about 87 points of OPS over six years, entirely hidden in the first table. The players who last the longest are not just healthy; they are the ones with the most room to fall and still stay in the lineup.

A second filter operates at the front of the curve. Players are only in the sample at 22 if they were good enough to reach the majors early, which makes the early rows look better than they would for the full population of prospects. The easy method therefore flatters both ends of the curve, which is why it looks like a plateau.

The delta method: following the same players

The delta method fixes this by comparing a player with himself. For each hitter who had 400 plate appearances in both a season and the next, it computes the change in OPS and averages the changes by age. Dropouts still bias the later rows, since a hitter must stay in the lineup two years running to count, but the comparison is no longer between different people.

Now the curve looks like what baseball people expect. Hitters improve through their early twenties, level off by about 26, and then decline a little more each year, with the average drop at 36 many times larger than the drop at 28. The turning point sits near 26 or 27, which is consistent with the old rule of thumb, associated with Bill James, that hitters peak around 27. Note how small the changes are, though: a typical loss of ten to twenty points of OPS a year in the early thirties is the difference between a good regular and a slightly less good one.

The noise is bigger than the signal

Averages hide spread. Among hitters aged 30 who played 400 or more plate appearances in both that year and the next, the standard deviation of the OPS change was about 0.086, while the average change was only -0.011. In other words, a typical player's year-to-year swing is roughly eight times the size of the average age effect. About 44 percent of those 30-year-olds actually improved the next year.

At 36 the picture is similar: an average decline of 0.026, a standard deviation of about 0.10, and 42 percent of hitters still posting a higher OPS the next season. The curve is real, but it is a tilt in a very noisy process. Part of that noise is ordinary luck. A hitter's results over 600 plate appearances move around for reasons no skill curve explains, which is the subject of sample size and regression to the mean. Part is injury, role changes and park changes.

This is why projection systems that use aging curves also regress hitters toward the average, and why a single player can sit well off the line without anything being wrong. The practical lesson: a 33-year-old hitter is a bit more likely to decline than to improve, and that is all the curve says.

How the skills age differently

OPS blends several skills, and they do not age at the same rate. The general findings in published research, which this article does not re-derive, are consistent: running speed declines earliest, which is why stolen bases and range at the demanding defensive positions drop first. Power tends to hold up longer than speed, and the ability to draw walks holds up longest of all, because it depends on judgment, not reflexes. This is why older hitters often keep a good on-base percentage while their batting average and doubles fall.

Two things follow. First, a player whose value is built on speed and defense has a steeper aging curve than a player whose value is power and patience. Second, the shape of the curve depends on the measure. The curve for stolen bases is much steeper than the curve for walks, and a single OPS curve averages them together.

For readers who like the newer measurements, the underlying contact skills are measured in exit velocity and related data, discussed in the guide to exit velocity, launch angle and barrels. The Lahman database does not carry those measures, so I will not quote them here, but they are a better window into whether a hitter is losing bat speed than a batting average can be.

Pitchers: strikeouts go first

Pitchers behave differently from hitters, and the data show it. Use pitchers with at least 100 innings in a season, from 1990 to 2025, and the easy method again gives a nearly flat line: average ERA is between 4.0 and 4.4 at nearly every age. The delta method shows what is underneath.

ERA rises on average from the late twenties onward, and strikeouts per nine innings fall from about 29, with the largest losses in the late thirties, though those rows hold few pitchers. The pattern fits the idea, again from general research and not from this table, that stuff fades before craft: a pitcher can learn to locate, mix pitches and sequence better, but the fastball and breaking-ball movement that produce strikeouts decline with age.

Two cautions. These ERA figures are not adjusted for the league's run environment, which swung over 36 seasons, or for ballpark, a factor explained in park factors. And the pairing requirement is harsher on pitchers than hitters, since a pitcher must remain healthy and effective enough to throw 100 innings in consecutive years. The pitchers who lost their jobs or their arms are missing from the pairs, so the true average decline is steeper than the table shows.

The exceptions that prove nothing

Every aging study invites a counterexample, and baseball has famous ones. Ted Williams hit .388 in 1957, a season in which he was 39 by the simple count, with 38 home runs in 420 at-bats. Nolan Ryan struck out 301 batters in 1989 at 42, after a career that began in 1966; his earned run average that year was 3.20. And Willie Mays, who hit 52 home runs in 1965 at 34, had 18 in 1971 and a .211 average in his last season in 1973 at 42.

Examples like these are useful for two reasons. They show the range, which is wide: Williams and Ryan stayed elite years longer than the curve suggests, while Mays shows the typical glide path of an all-time great. And they illustrate selection. We remember Williams and Ryan because they were the ones who defied the average, which is the same survivorship effect in narrative form. For every pitcher who struck out 300 at 42 there are many more whose careers ended earlier and left no memorable season behind.

One honest caution is that the period from 1990 to 2005 included performance-enhancing drugs, and any curve built from those years may include seasons affected by them. Players were suspended and tested positive, and some late-career surges drew suspicion. Nothing in the Lahman database can separate those seasons, so a curve like this one measures what happened, not what would have happened without them.

What changes the curve

Several things move the curve for a given player or era. Position matters in principle, since the more athletic positions ask more of the body, and so does a player's mix of skills, because speed-dependent players age faster. Era matters: the league's run environment, the strike zone and the rules for the ball and the bases change what a good season looks like. Training and medicine may matter too, though the data here do not test that.

Those factors explain why published curves differ, with some finding a peak at 26 and others at 29. The disagreement is usually less about the data than about the method: age definition, playing-time cutoff, which measure and how the survivors are handled. When you see a claim about peak age, ask whether it uses the delta method and what happens to players who drop out.

How to use this without being fooled

For fantasy owners, the useful conclusions are modest. Prefer younger players among equal projections, because the expected change favors them: a 24-year-old is expected to improve and a 34-year-old to decline, other things being equal. Treat speed as perishable. And expect wide ranges, since the spread around the average is far bigger than the average itself. The guide to expected statistics is a good companion for judging whether a hitter's results match his contact quality.

For anyone reading a projection, the question to ask is what the system assumes about age and whether it regresses toward the mean. A reasonable projection for a 35-year-old subtracts a little for age and does not pretend to know which of the many possible outcomes will occur. You can test your own intuition with the stat calculator and check the career leaders on the leaders page.

Common questions

At what age do hitters peak? On the delta method used here, the average hitter stops improving around 26 or 27, which matches the common rule of thumb that peak years run from about 26 to 29. Individual players vary widely.

Do pitchers peak earlier or later than hitters? In this data, pitchers' strikeout rates begin to slip in their late twenties, so the arm peaks early, while ERA changes are small until the early thirties. Command and sequencing let many pitchers stay effective for longer than their strikeout rates would suggest.

Why not just average every player at each age? Because the players who remain at older ages are survivors. The easy average blends different people at each age, which is why it hides decline.

Are players aging later than they used to? Many analysts think so, but this article does not test it. The numbers above cover only 1990 to 2025 and mix the whole period together.

Can an aging curve tell me whether one player will decline? No. It gives the average tilt for a group. The spread around that average is larger than the average itself, so treat it as a prior and not a forecast.

Where to go next

Start with sample size and regression to the mean, then read about park factors, which matter for any comparison across seasons. For more analysis on the site, browse the stats and analytics topic page.