Yesterday's picks: how the model did (July 23)

Thursday was a split decision for Verdexed's MLB model. The single side it was most confident in cashed, and a couple of the sharper value reads landed, but the middle of the card was uneven: two of the day's moderate home favorites got buried outright. On balance it graded out as a roughly even day, the kind of result a market-anchored model produces far more often than a clean sweep in either direction.

This column exists to grade the model in public, win or lose. Thursday was neither a banner day nor a disaster, and the honest read is that it landed somewhere in the middle.

The scorecard

The defining feature of the slate was the top of the confidence ladder holding while the tier just below it wobbled. Verdexed's highest-conviction side of the day was Atlanta, which the model gave a 68.7% win probability at home against San Diego. That one held. It was the strongest single number on the board, and it delivered.

Below that marquee lean, the picture got messier. Two of the day's clearest moneyline favorites did not hold. Cleveland sat around -145 at home behind Gavin Williams, with roughly two-thirds of the public moneyline tickets on the Guardians, and lost. St. Louis was a home favorite in the -130 range against Arizona and lost by a wide margin. When the model leans with a favorite that price implies should win close to 58 to 60 percent of the time, those are the games it expects to bank most nights. On Thursday, both of those went the other way.

Working against that, a couple of the model's contrarian and value-side reads cashed. Toronto, the home side that drew the sharper money at a near even-number price against Tampa Bay, won behind a strong Shane Bieber start. Houston, favored at home against Miami, took care of business. So the losses in the mid-tier were partly offset by the sides where the model sat with, or slightly against, the market and got it right.

A precise graded record for the day, including the exact hit rate and the Brier score, was not retrievable from Verdexed's public accuracy surface at the time of writing. Qualitatively, this reads as a near coin-flip day: the best number on the board held, the flagged value spots mostly cashed, and the moderate favorites split with two notable ones falling. That is a fair distance from a green day, but well short of a red one.

Best call / worst miss

The best call was Atlanta. The model's top-rated side of the slate at 68.7%, the Braves beat the Padres 6-5, and they got there in the style you want from a high-confidence pick: Chris Sale matched a season high with 11 strikeouts and Atlanta hit four home runs. When the single firmest number on the card is also the game that plays out cleanly, that is the model doing exactly what it is supposed to do at the top of the ladder.

The worst miss was Cleveland. The Guardians were a moderate home favorite around -145, the kind of price the model treats as a lean rather than a lock, and they lost 3-1 to the Twins. Minnesota's Taj Bradley, carrying a strong track record against Cleveland, shut the favored bats down on the road. It is the sort of game the model expects to win the clear majority of the time, and it did not. St. Louis was close behind on the miss list, dropping a 10-6 decision to Arizona as a home favorite in a game that was expected to be a coin flip and turned into a route.

There is no spinning either of those. A favored home side that gets outpitched and a favored home side that gets outslugged are both simply losses, and they are the reason Thursday grades as a split rather than a clean win.

What it means

One uneven Thursday is noise, not a signal. The tell to watch over a longer window is not whether every favorite holds on a given night, which no honest model expects, but whether the top of the confidence ladder keeps holding at roughly the rate the model advertises. On Thursday it did: the 68.7% pick cashed. The soft spot was the middle tier, where two moderate favorites fell in the same day, and that is exactly the band where variance shows up most because those games are the closest to true toss-ups once the price is stripped out.

The model is market-anchored by design, which means its edge on any single game is thin and its day-to-day results should cluster near even more often than they blow out in either direction. A split like Thursday, with the marquee lean holding and the mid-tier chalk taking a couple of losses, is squarely inside the normal range of outcomes. It is worth logging, not overreacting to. The number that matters is the season-long calibration, and one middling day does not move it much in either direction.

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.