What this page actually compares
Two careers, three ways, in that order down the page. First a six-sided shape for each man, drawn from percentiles rather than raw numbers. Then the rate statistics, which can be read straight across. Then the totals, which cannot — and the page says so at the point where you would otherwise be misled, rather than in a footnote nobody reaches.
Everything is assembled in your browser out of files fetched from this site the moment you pick a name — one slice of the career data per surname, and the page tells you how large it is while it is coming. Most of them are small; the commonest opening letters are not, because a great many ballplayers share them. Once a name is loaded it stays loaded, so swapping the other man out, flipping between hitting and pitching, or reading the same pair on the age axis instead of the calendar all happen without another download — and the address bar keeps the pair, which means the comparison you are looking at is a link you can send someone.
The six axes, and what they are measured against
A hexagon is the easiest chart in the world to lie with, because six decisions disappear into one pleasing shape. So here they are. Each axis is a percentile against the players in the same league in the same season — the men he actually played against — and a career figure is the average of his seasons weighted by how much he played in each of them. The real number sits next to the percentile in the table under every chart, because a percentile on its own is a number with its assumptions hidden.
Hitters are placed on Contact, Power, Eye, Speed, Glove, Durability. Pitchers are placed on Strikeouts, Control, Home runs, Efficiency, Prevention, Workload. Four of the pitching axes measure something you want less of — walks, home runs, baserunners, runs — and those are flipped before they are drawn, so that on every axis of every chart, further out is better.
Why not one percentile across all of history? Because it would flatten the dead-ball era into nothing. Seven home runs in 1908 made you a slugger; the same seven in 1999 made you a shortstop having a quiet year. A single all-time distribution would push every batter before 1920 to the bottom of the power axis and call it a finding, when all it would show is that the ball changed. Comparing a man with his own contemporaries is the only thing that travels across a century and a half, and it is the same logic the WAR figures on this site already use.
The fielding axis is the weak one and we would rather say so. Our fielding estimate is built from putouts and assists, because that is what the source has; there is no range data and no fielding innings at all before 1954. The result understates the very best defenders — the shape will show a famous glove as merely good. Treat that axis with more doubt than the other five, and treat it as the reason the fielding component of our WAR is the part we trust least.
Why the totals are not converted
Seasons have not been the same length. A Negro Leagues club in 1937 played about 51 league games while an American or National League club played about 155. The 1876 National League played about 70. The 1918, 1919, 1981, 1994 and 2020 seasons were all cut short for reasons that had nothing to do with the players in them. Put two career home run totals next to each other from either side of one of those gaps and the screen has told you something true and left you believing something false.
The fix people expect is to scale the shorter seasons up to a modern schedule. We do not do that, and it is a deliberate refusal. The totals on this site add the seasons exactly as they were played, which is why 714, 755, 511, 5,714 and 4,189 come out matching the public record — and the moment we start pro-rating, that agreement is gone and we are quoting numbers we invented. So instead the page puts the facts next to the totals: how many seasons, how many games, how many games in an average season, and which leagues. When the two men played schedules of noticeably different length, it says so in a sentence directly above the table rather than leaving you to notice.
The rate statistics are a different matter. A batting average, an on-base percentage, an earned run average — none of them care how many games were on the calendar. Those are in their own table, with a heading that says they compare straight across, because the useful distinction here is not between hitting and pitching but between numbers that survive the comparison and numbers that need a caption.
Two-way players, and comparing a hitter with a pitcher
A handful of men did both jobs well enough to have two shapes, and they are the most interesting thing in this data. When a player has enough hitting seasons and enough pitching seasons, both hexagons are drawn and both sets of totals appear; the page does not pick one job and quietly drop the other.
Comparing a hitter with a pitcher has no clean answer, because the two shapes do not share a single axis. There is no honest way to overlay strikeouts per nine innings on isolated power. So the boxes above will not offer you the pair in the first place: put a hitter in one and the other only lists hitters, and it says on screen that it has done so and why. Showing somebody a road and then explaining why it was a dead end is worse than showing only the roads that go somewhere. An address you were given with one of each in it still opens — that is what the note at the top of such a comparison is for — and the two men who did both jobs are handled below.
Two other things the search quietly does. Anyone whose record has neither a plate appearance nor an inning — a hundred and fifty-one men who appeared in a game and did neither — is left out of these boxes, because there is nothing on either chart to put him on. And when a man did both jobs, he shows up in whichever list you are filling: Babe Ruth is offered as a pitcher when you are choosing an opponent for Greg Maddux and as a hitter when you are choosing one for Tony Gwynn.
8,632 players have enough hitting seasons to be placed on the batting hexagon and 8,213 on the pitching one; 920 of them appear on both. Everybody else is still in here with full career totals — a shape needs a hundred plate appearances or twenty innings in a season before a percentile means anything, and below that line a single good afternoon would put a man in the ninety-ninth percentile.
When both men did both jobs — Ohtani against Ruth is the pair everyone wants — the page cannot know which half of the argument you are having, so it asks. A pair of buttons appears under the names, the answer decides which chart is at the top, and it goes into the address with the two identifiers, so the link you send opens on the same half you were looking at. Neither career is dropped: the other job is still further down the page in full. When only one of the two did both jobs there is nothing to ask, because the answer is whichever job the other man did, and the buttons stay away.
What the address bar is for
Picking two names rewrites the address, and opening that address gives the same two names back. That is the whole point of the feature: an argument about two players is not worth much if you cannot hand it to the person you are arguing with. The two identifiers in the address are the entire state of this page, so a bookmark reopens the same comparison months later and a link drops the other person straight into it with both names already picked.
Each name also links through to its own career page, where every season is listed in full with league-by-league subtotals. Start from a season, from a player, or from the page that explains which of those you want.