Yesterday's picks: how the model did (July 29)

Wednesday was a losing day for Verdexed's MLB model, and it lost at both ends of its own card. The column published five reads on the July 29 slate. Three carried a stated direction and one landed. The firmest side lost 6-5 in 12 innings, the widest gap between price and inputs lost 4-2, and the day's one correct call won by 13 runs. A model that misses its chalk and its value read on the same night has had a bad night, and no framing improves it.

The scorecard

Three graded reads, one correct, two wrong. That is a 1-2 day.

The more damaging half is where the losses sat. The high-confidence tier, ungraded for two straight days after rainouts in New York, got its live test Wednesday and failed it: New York behind Cam Schlittler, listed near -147 and named the firmest side on a 15-game board, lost in Chicago. The widest disagreement on the board, the read against Los Angeles at a reported -270, lost as well.

The two remaining reads carried no side by design. Braves at Mets was written as an argument against paying a heavy Atlanta price rather than an endorsement of New York, and Cubs at Cardinals was published as a game with no gap wide enough to headline. Neither grades, leaving a graded sample of three.

A precise slate-wide accuracy percentage and Brier figure for Wednesday were not retrievable from Verdexed's public accuracy surface at the time of writing, so no number is claimed here. The verifiable material 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 San Francisco, and it was the read the column was least emphatic about. The argument was narrow: Milwaukee had been listed near -154 with the Giants around +130, Logan Webb's 5-7 record understated him, and a public projection had the game at 50.8 percent for San Francisco, essentially a pick'em with a road side laying a real number. The stated lean was toward the home underdog on price, not on any claim San Francisco was the better team. The Giants routed the Brewers 16-3, Webb allowing two runs over six innings to move to 6-0 in his career against Milwaukee. A pick'em call that wins by 13 is still a pick'em call, and the margin is not evidence the model saw more than it said it did.

The worst miss was Los Angeles, worse than the scoreline suggests because the model's stated mechanism held and the pick lost anyway. The read was that a 73 percent implied number was too much to ask of a club carrying Blake Snell, Tyler Glasnow, and Will Smith on the 60-day injured list, with the starter unconfirmed. The starter question resolved against the Dodgers on paper: Eric Lauer, 5-5 with a 4.83 ERA, took the ball opposite Seattle's Emerson Hancock and his 3.16. Los Angeles won 4-2 regardless. Andy Pages punished an intentional walk to Shohei Ohtani with an RBI single in the fourth, and Edwin Diaz closed it out in his first game back. When the argument is right about the inputs and the market is right about the outcome, the market wins the day and the model absorbs the loss.

The Yankees loss belongs in the same section. Schlittler was strong into the seventh and Spencer Jones homered for three early, but New York's 4-1 eighth-inning lead vanished on consecutive homers by Sam Antonacci and Munetaka Murakami off Paul Blackburn. Anthony Volpe's pinch-hit sacrifice fly put New York back ahead in the 11th, and Edgar Quero singled home Tristan Peters in the 12th for a 6-5 Chicago win. The column's stated caveat was that run support is outside a pitcher's control and a 7.5 total leaves one swing deciding it. The caveat was right and the pick was wrong, which counts as a loss.

Ungraded context worth recording: the Atlanta side this column flagged as its firmest read Monday, then reported washed out twice, finally played as Wednesday's nightcap, and Chris Sale struck out nine over two-hit ball in a 1-0 Braves win. Atlanta dropped the opener 3-2 behind Sean Manaea to split the doubleheader, the direction the day's no-side read had pointed. Neither result appears in Wednesday's record, because neither was a stated July 29 pick.

What it means

One 1-2 day on three reads is a bad day and not much more than that. Tuesday's card went 3-0, which puts this column's own published sides at 4-2 across the two days, and neither number is a trend. The model's calibration is settled on the accuracy page over weeks rather than in a column written the next morning.

What carries forward is narrower and less comfortable. The high-confidence tier had its first live test since Sunday and lost it. That does not make the tier broken, since a well-calibrated 60 percent side loses four times in ten by design, but the tier's only data point this week is a loss. The honest summary is that the model produced very little gradeable evidence in the last three days, and what little it produced Wednesday went against it.

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.