Yesterday's picks: how the model did (July 27)

Monday was a losing day for Verdexed's MLB model. The column published five reads on the July 27 slate, three of them carried a stated direction, and the model went 1-2 on them. The lean on Washington, described in advance as the widest gap between price and projection on the board, lost 3-2 to Toronto. The lean on Milwaukee lost 3-0 in San Francisco. Chicago, the read built on the argument that a stale season ERA was mispricing a starter, won 7-3 in St. Louis. The day's firmest side never threw a pitch.

The scorecard

Two of the five reads do not grade, for two different reasons.

Guardians at Reds, the firmest side on the board, was postponed by rain in Cincinnati and rescheduled as a day-night doubleheader Tuesday, with Chase Burns moved to Game 1. Mariners at Rangers was written as the clearest spot where the market looked slightly ahead of the projection rather than behind it, and no side was endorsed.

That leaves three graded reads and a 1-2 record. It also means the high-confidence tier went ungraded entirely on Monday. The only stated high-conviction side, Cincinnati at home behind Burns, was washed out, so the day's record is drawn from the middle and the value end of the card and nowhere else. Readers should weigh it accordingly: this is not a test of the model's chalk, it is a test of its edges, and its edges lost.

A precise full-slate graded record for Monday, including the day's accuracy percentage and Brier score, was not retrievable from Verdexed's public accuracy surface at the time of writing, so no figure is claimed here. What is verifiable is the published output and the final scores, and on that basis Monday was one for three.

Best call / worst miss

The best call was Chicago, and the mechanism behind it worked exactly as written. The column argued that David Peterson's 6.45 season ERA was the kind of stale input a market can lag and a recency-weighted model can catch, pointing to his much stronger work since arriving from the Mets. Peterson went 5 2/3 innings at Busch Stadium and was charged with two runs, one of which scored after he left the game. Alex Bregman and Nico Hoerner each collected four hits. The Cubs won 7-3. The argument and the outcome matched, which is the only kind of correct call worth much.

The worst miss was Washington, and it was the day's headline read. The entire case was that the market near -130 was pricing Max Scherzer close to a normal starter and should not have been, given six weeks lost to back spasms, three cortisone injections in his right thumb, and an opponent leading the majors in home runs. Scherzer, starting on his 42nd birthday in his first big league appearance since June 10, allowed one run on four hits over 2 2/3 innings on 75 pitches. The capped pitch count arrived precisely as predicted, and it did not matter, because Toronto's bullpen finished the job. Luis Urias homered, Myles Straw doubled and drove in a run, and the Blue Jays won 3-2. Washington, the highest-scoring offense in the sport, scored twice. The model correctly identified the risk that would decide the game and still finished on the wrong side of it.

The Milwaukee loss was simpler and no better. With the listed starters close to a wash, the model was left leaning on roster strength and bullpen depth alone. Tyler Mahle threw six sharp innings, three San Francisco relievers finished the shutout, and Grant McCray hit a two-run homer in the fourth, his first of the season. Giants 3-0. A 65-39 team does not beat a 44-60 team because of the standings.

One ungraded note belongs on the page. The caution on Seattle looked right. Texas won 7-3 and took three of four, and George Kirby gave up four home runs inside the first three innings, two of them to Joc Pederson. Declining to endorse a number is not a win and does not go on the record.

What it means

One Monday at 1-2 on three graded reads carries almost no signal by itself. The three-day frame is more useful and less dramatic: Saturday split its directional reads, Sunday went 2-1, Monday went 1-2. That is ordinary output across a board of thin edges, not a slump.

The narrower item is the one worth watching, and it is now the second straight day this column has had to write it. Sunday's single widest listed edge, Cleveland as a road underdog in Tampa Bay, was shut out 1-0. Monday's single widest listed edge, Washington against a compromised Scherzer, lost 3-2. Both lost on the specific mechanism the model used to justify them, which is a worse outcome than losing to noise. Two days is not a pattern and does not establish one. If value reads keep losing on their own logic across a longer run of days while the confidence tier holds, that becomes a calibration question rather than a variance question. Monday could not contribute to the other half of that comparison, because the rain took it. The accuracy page settles this over weeks, not this column over days.

Responsible play

**This is a record of past model output, not betting advice.** Verdexed grades its model in public for transparency and entertainment. Past results do not predict future outcomes, nothing here is a recommendation to place a wager, and no pick is ever a guarantee, even a strong model loses regularly.

**You must be 21+** (or the legal age in your jurisdiction) and located where sports wagering is legal to bet. Sports betting is not available everywhere; know and follow the laws where you are. Never bet more than you can afford to lose, and never chase losses.

If gambling is affecting you or someone you know, help is available. In the US, call or text **1-800-GAMBLER** (1-800-426-2537), or visit ncpgambling.org. Please play responsibly.