Yesterday's picks: how the model did (July 30)

Thursday was the second straight losing day for Verdexed's MLB model. The column published reads on seven games from the July 30 board, three of which carried a stated direction. One landed. The firmest side won on a night its central argument fell apart three innings in, the widest gap between price and inputs was buried by the exact pitcher the model doubted, and a soft home lean lost by three. That is a 1-2 day, and the way the win arrived does not make it look better.

The scorecard

Three graded reads, one correct, two wrong.

The correct one was Tampa Bay behind Shane McClanahan against Texas, named the day's firmest side on a 10-game board. The Rays won 3-2. The two losses were the read against Los Angeles at a reported -196 versus Seattle, published as the widest disagreement on the board, and a modest lean toward the home side in Giants at Padres. Los Angeles won 6-2. San Francisco won 4-1.

The rest of the card carried no direction by design and does not grade. Marlins at Mets was written as an argument against paying a road price on Miami rather than an endorsement of New York, and Nationals at Braves, Cubs at Cardinals, and Red Sox at Athletics were published with unconfirmed probables, which the column called an absence of a lean rather than a lean.

A slate-wide accuracy percentage and Brier figure for Thursday were not retrievable from Verdexed's public accuracy surface at the time of writing, so no number is claimed here. What is verifiable is the published output and the confirmed final scores, and on that basis the model went one for three.

Best call / worst miss

The best call was Tampa Bay, and it is the most honest entry on the card because the model was right about the team and wrong about the reason. The case rested on McClanahan, 9-6 with a 3.09 ERA and two runs or fewer allowed in five of his last six starts, supported by a 39-16 home record against a Texas club that had been 27-28 on the road. McClanahan lasted three innings and left with tightness in his back. Tampa Bay won anyway, 3-2, on Jonny DeLuca's home run and Cedric Mullins' go-ahead shot in the eighth, with seven pitchers combining on a two-hitter. The Rays moved to 64-44. The pick cashed. The mechanism behind it did not survive the fourth inning, and a model that leaned on a starter's form got bailed out by a bullpen it never mentioned.

The worst miss was Los Angeles, and it was clean. The read was that a -196 number, implying roughly 66 percent before the sportsbook's margin, asked too much of a rotation starting Roki Sasaki and his 5.33 first-half ERA, with Blake Snell still on the 60-day injured list and Shohei Ohtani unavailable to pitch. The column flagged Sasaki as projected rather than confirmed. He started, struck out seven over 5 1/3 innings, allowed two hits through five, and lowered his ERA to 4.64, its lowest mark since early June. Dalton Rushing's three-run homer and Teoscar Hernandez's solo shot did the rest in a 6-2 Dodgers win. The model built its widest disagreement of the day on one pitcher's ERA, that pitcher pitched the best start of his second half, and the price was correct.

The Padres loss is smaller but belongs in the record. The lean was qualitative, toward the home side on park and matchup, with no market number retrievable and none quoted. Robbie Ray held San Diego to a run on four hits over six innings, rookie Daniel Susac homered and drove in three, and San Francisco won 4-1. A soft lean is still a lean, and it lost.

Ungraded but worth recording: the no-side read on Marlins at Mets pointed away from Miami's road price, and New York won 4-2 behind Nolan McLean. The direction was right, but it was not a stated pick, and counting it would be the kind of retroactive credit this column exists to avoid.

What it means

Two straight 1-2 days is a bad stretch and not yet a signal. Tuesday's card went 3-0, which puts this column's own published sides at 5-4 across three days. Nine graded reads is a sample that tells almost nothing about calibration, and the accuracy page settles that question over weeks rather than in a morning column.

The more useful observation is about where the losses are clustering. Both of the last two days lost the widest-edge read, and both times the loss came from the model disagreeing with a market price on a starting pitcher. Wednesday it doubted a Dodgers number with Eric Lauer on the mound and lost. Thursday it doubted a Dodgers number with Sasaki on the mound and lost again. One club, two prices, two arguments that the market was overrating Los Angeles, both wrong. That is either variance or a read the model is getting backwards, and two data points cannot separate the two. It is worth watching rather than concluding.

The high-confidence side won Thursday, which is the one piece of good news, and it won for reasons the model did not identify.

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.