Bullpen Report, August 6

20 save sources ranked by recent workload, through 2026-08-04.

20 pitchers with a real claim on save chances, ranked by workload. The usage window ends 2026-08-04, so last night is not in the data yet and this is a workload report rather than a call on tonight, saying so is more useful than implying otherwise.

Worked hardest

The most pitches thrown across the window. Volume is the leading indicator of a scheduled day off, and of the committee that tends to follow one.

Freshest arms

The lightest workloads among pitchers who still hold a claim on saves.

3 of these pitchers appeared at least twice inside a three-day window. That is the pattern worth tracking: back-to-back usage is what turns a clean closer job into a committee for a night.

What this means

Bednar is the only genuinely tired arm on this board. Two appearances and 46 pitches inside three days, and he is sitting on two days rest, which means the Yankees have already given him the breather. That is a back-to-back-plus cleanup workload, and 46 pitches is more than double what most of the names above threw in the same window. If you are streaming a second closer tonight and your only tiebreaker is availability, Bednar is the one I would trust least to finish a clean inning, and the one most likely to get the ninth anyway because the alternative is worse. Nothing here says his job is in danger. It says his next outing is likelier to be scruffy.

The number that should actually change a decision is Clayton Beeter's. Ten saves banked, 22 projected the rest of the way, on eight pitches in three days. That is a projection saying he converts more saves from here than he has all year, which only happens if the model thinks he has just taken the job outright rather than shared it. The mechanism is not visible in this feed: we can see the workload is light and the forward number is heavy, but nothing here tells us whether that is a role change, a schedule loaded with close games, or both. The honest read is that the engine has repriced his role and this table does not show you why.

The case against Beeter is straightforward, and it is a good one. Eight pitches in three days is exactly what a setup man with one low-leverage inning looks like, and the same feed shows Bryan Baker, on 13 pitches, sitting on 26 season saves with 30 projected. Baker's forward number is supported by his back number. Beeter's is not supported by anything you can see. If a projection has to more than double a pitcher's save rate to be right, the burden is on the projection. What settles it is the next week: two save chances taken with a clean handoff and the role is real, and 22 is live. One appearance in the seventh inning of a four-run game and the number is stale. Add him now if he is free, because the cost of being wrong is a bench spot and the cost of being late is 22 saves.

Sewald runs the other direction, and he is the one I would move. Eighteen saves already, 12 projected the rest of the way. Nothing in his workload explains it: one appearance, 19 pitches, two days rest, all perfectly normal. A forward number below the back number on a healthy arm at this point in the season is the model pricing in a smaller share of Arizona's ninth innings, or fewer of them. Sell him into the name recognition his 18 saves buy you, and do it before somebody in your league notices the projection.

The rest is quiet, and I would say so plainly rather than manufacture a story. Latz at 21 and 22 projected is the picture of a stable job. Iglesias, Duran and Scott all pair modest recent usage with forward numbers that sit right on top of their season pace. Megill's jump from 15 to 26 is real but it reads as a healthy full-season workload rather than a role change, and Milwaukee has given him ten pitches in three days. Nobody in that group needs an action tonight. Beeter and Sewald do.

How to read this

Rest is counted from the dates a pitcher actually appeared, not from a "pitched yesterday" flag. That distinction matters: such a flag is relative to when the data was built, so on any day the feed has not refreshed it quietly shifts by one and reports a pitcher as having thrown last night when he threw the night before. Dates cannot drift that way.

The window is three days, so "3+ days" is a floor rather than a measurement, it means no appearance in the window, which is everything the source knows.

Only pitchers with a genuine claim on saves appear here, measured by the same closer-confidence score the depth charts use. A workload report covering every reliever would be longer and would answer nothing.

All articles · Redraft rankings · Dynasty rankings