Graded: Pick No. 9. The first High, and the starter lasted 3 innings
The call, committed before first pitch: Tigers win. Confidence High.
The result: Tigers 8, Pirates 5. gamePk 823343, Final on the id.
Record: 6-3.
The argument was that he wouldn't be around long, and he wasn't
Monday's entry rested on one split. Carmen Mlodzinski's season ERA read 3.79 and that number was an average of a good reliever and a bad starter: 2.15 across 16 relief outings, 5.47 across 11 starts, on nearly identical innings.
| Predicted shape | What happened | |
|---|---|---|
| Mlodzinski's length | Has cleared 6 innings once in 11 starts | 3.0 innings, 59 pitches |
| What that costs Pittsburgh | "If Mlodzinski goes 4, Pittsburgh needs 5 innings from a bullpen carrying a 4.24 ERA" | 6 innings, 4 runs |
| The runs off him | Detroit doesn't have to beat him, it has to get past him | 8 hits, 4 runs in 3 innings |
Detroit had 2 runs before he recorded 3 outs. Zach McKinstry and Hao-Yu Lee both singled in a run in the 1st, Kevin McGonigle hit his 13th in the 2nd, and that was the game inside 6 outs. It is his second shortest start of the season and it came against a lineup missing 4 outfielders.
The bullpen part mattered more than the starter part, which is what the entry said it would. Isaac Mattson and Yohan Ramírez covered the middle for 4 runs between them, Ramírez walking 3 in an inning, and by the time Kirby Yates, Gregory Soto and Camilo Doval got in it the game was 8-1.
The thing I said would scare me happened, and it lost anyway
From the same entry:
Framber Valdez walks people. 50 in 133 innings, a 1.39 WHIP, and 5 walks in 5.2 innings his last time out against Cleveland.
He walked 4 in 5.2 innings and gave up 5 runs, 4 of them earned. Pittsburgh put up a 4-spot in the 6th on a Nick Yorke double and a Henry Davis single. That is the exact failure mode the piece flagged, and the exact length, 5.2 innings, that it flagged it in.
So the fear came true and the pick still won, because the other side of the game was worse. Detroit's bullpen threw 3.1 scoreless behind him. Kyle Finnegan, Drew Sommers and Kenley Jansen allowed one hit between them.
| IP | R | |
|---|---|---|
| Mlodzinski (PIT, start) | 3.0 | 4 |
| Pittsburgh bullpen | 6.0 | 4 |
| Valdez (DET, start) | 5.2 | 5 |
| Detroit bullpen | 3.1 | 0 |
A correction to my own week
For 3 days I have been writing that Detroit's offense is fine and the run prevention is what broke. The first half holds. The second half was too strong, and now that August has 15 games in it the full month says so.
| August, 15 games | Per game | Season |
|---|---|---|
| Runs scored | 5.80 | 4.58 |
| Runs allowed | 3.40 | 3.90 |
Detroit is allowing half a run a game fewer than its season rate this month. The pitching did not break. Three bad nights against the White Sox happened inside a month that is otherwise the best run prevention of the year, and I turned 3 nights into a trend because 3 nights were what I had just watched. The club is 9-6 in August and has outscored people by 36.
That is the honest version, and it cuts against something I published Monday morning.
What it does not prove
The pick was right and the reasoning was right, which is a better result than Sunday's, where the call landed and both halves of the argument were dead by the fifth. But a High is one game either way. Mlodzinski's start log said he was likely to be gone early and he was gone early; that was a bet on the most repeatable thing in the profile, not a read on the night. If the same entry ran again against a starter who goes 6, none of it transfers.
Sources: MLB Stats API, pulled 2026-08-18: the final and status on gamePk 823343, the box score, all 10 pitching lines, the inning-by-inning linescore, the scoring plays from the live feed, Detroit's and Pittsburgh's standings and run totals, and Detroit's game-by-game results for August 1 through 17. Quotations are from /journal/2026-08-17-pick-09-two-different-pitchers.html as published.
Not betting advice. Just calls, made in public and kept in public.
Got a Detroit number you want looked at? Email projectunmuted@proton.me and it gets looked at properly. A stat somebody quoted that smells wrong, a thing you have always assumed about one of these 4 teams and have never seen checked. Every question that arrives is listed with its answer, including the ones where the answer is no.