The hold is relief pitching’s emptiest stat
The hold exists because the save left a gap. Once the save turned the ninth-inning reliever into a celebrated, well-paid specialist, the setup men who bridged the seventh and eighth had no comparable number to point to at contract time. So the hold was invented to fill that void: a credit for a relief pitcher who enters in a save situation, records at least one out, and leaves with his team’s lead intact. It gave setup relievers a countable badge and a way to say they, too, protected leads. But the hold inherited every weakness of the save and added a few of its own. It rewards being summoned in the right situation more than pitching well in it, it can be earned on a night a reliever pitches badly, and it tells you almost nothing about whether the pitcher who got it is any good.
What a hold actually records
A hold is awarded when a relief pitcher enters in a save situation, gets at least one out, and hands the game off without surrendering the lead or finishing it. That is the whole definition, and notice what it turns on: the score when the manager happened to call your name. A pitcher who enters with a three-run lead, allows two runs, and departs still ahead gets a hold. A pitcher who throws two perfect innings in a tie game gets nothing, because a tie is not a save situation. The hold is triggered by the game state at entry and the game state at exit — a bookkeeping condition — not by the quality of the pitching in between. Like the save, it is a stat about the circumstance a reliever was dropped into as much as anything he did once he got there.
You can pitch badly and still get one
Because the only requirements are an out and a preserved lead, the hold is remarkably forgiving. Enter with a comfortable cushion, allow a couple of hits and a run, get a single out, and leave still in front, and the line reads “hold” exactly the same as a spotless, high-leverage inning against the heart of the order. The stat cannot tell the difference between a reliever who slammed the door and one who nearly let the house burn down but was bailed out by a big enough margin. It records that the lead survived his appearance, not that he protected it well, and those are very different claims. A number that a pitcher can earn while actively making his team sweat is not measuring effectiveness.
Leverage is invisible to it
The hold flattens the single most important variable in relief pitching: how much the game was on the line. A reliever who protects a one-run lead with the bases loaded and the opponent’s best hitter up is doing something enormously more valuable than one nursing a three-run lead against the bottom of the order, yet both collect one hold. Managers deploy their best relievers in the highest-leverage spots, but those spots are not always ninth-inning save chances or tidy setup situations, and the hold only sees a narrow slice of them. Because it ignores leverage entirely, the stat lumps the genuinely difficult work in with the routine, and a season total of holds tells you how often a pitcher was used in qualifying spots, not how much he actually helped win games.
It is a usage stat in disguise
The deepest problem is that hold totals are driven by role and opportunity. A pitcher slotted as the primary eighth-inning setup man on a good team — a club that plays a lot of close games it leads late — will rack up holds simply by being handed the right innings again and again. Move that same pitcher to a bad team, or to a middle-relief role, or to an offense that wins in blowouts, and his hold total collapses without his ability changing at all. This is the classic signature of a context stat: the number swings on the manager’s bullpen chart and the team’s record, not the arm. Ranking relievers by holds ends up ranking who got the setup job on a contender, which is a fact about deployment, not a measure of talent. The same distortion shows up in the way holds are chased at the trade deadline, where a reliever’s hold total gets quoted as evidence of late-inning pedigree when it often just reflects that he pitched the eighth for a team that was frequently ahead. Strip away the role and the record of the club around him and the number has little left to say about the pitcher himself.
Where the hold has any use
The hold is not entirely without information. As a rough marker of role, it tells you a pitcher was trusted with leads in the late-middle innings, and over time a steady stream of holds signals that a manager kept going back to him in those spots. It also captures, loosely, that a reliever was mostly successful at the narrow task of not blowing the leads he inherited. Read as “this pitcher was the setup guy and generally held serve,” the stat is harmless shorthand. The mistake is treating it as a performance metric — using hold totals to argue one reliever was better than another, or to set his market value — when it was never built to measure how well anyone pitched.
What to read instead
Evaluate relievers with numbers that actually track their pitching and the stakes of it. Strikeout and walk rates, and the fielding-independent metrics built on them, describe what a reliever controls. Run-prevention stats like ERA and its estimators, read with the small-sample caution relief innings demand, describe results. And leverage-aware measures — win probability added, and stats that weight innings by how tight the game was — capture the thing the hold throws away: whether a pitcher succeeded when it mattered most. Together those tell you whether a reliever is genuinely good and how much his work moved the needle. The hold answers a narrower, emptier question: was he standing on the mound, in a lead, in the right inning.
The honest read
The hold solved a fairness problem — setup men deserved some recognition the save denied them — but it solved it by copying the save’s flaws instead of fixing them. It rewards the circumstance of entry over the quality of the outing, can be earned on a shaky night, ignores leverage, and ultimately counts how often a pitcher was handed the right role on the right team. Use it, if at all, as loose shorthand for “late-inning setup role,” and turn to strikeout and walk rates, run prevention, and leverage-weighted value to judge whether a reliever is actually good. A pitcher gets a hold for keeping a lead alive through his appearance. Whether he pitched well enough to deserve the credit is a question the stat never asks.