Yesterday's picks: how the model did as its lone post-break lean fell in Philadelphia

Baseball came back from the All-Star break with a single game, and Verdexed's model came away 0-for-1. The lone contest on the July 16 slate was the Mets at the Phillies in Philadelphia, the marquee national opener before the full schedule returned a day later. The model's read on that game leaned toward the home favorite, and the home favorite lost. New York beat Philadelphia 4-1, so the day's graded record is a clean miss on a sample of exactly one pick.

That is worth stating plainly up front. A one-game day is the smallest possible slate, and losing it is not the same signal as losing a fifteen-game Saturday. But a miss is a miss, and the honest line is that the only side the model graded on Thursday did not cash.

The scorecard

One game, one graded pick, zero correct. Hit-rate on the day: 0 percent.

The Phillies opened and closed as the home favorite, priced in the -124 to -135 range across the market, which implies roughly a 55 to 58 percent chance to win. That is a lean, not a hammer. There was no high-confidence tier to speak of on July 16 because there was only one game on the board, and it was a modest favorite rather than anything the model would have flagged at the 65 percent-plus or 70 percent-plus levels it reserves for its strongest reads. So the tier breakdown that usually carries these recaps has nothing to grade: no premium chalk held, and none fell, because none existed. What existed was a single near-coin-flip that tipped the wrong way.

On the calibration side, a single mispriced outcome does little to a Brier score built on hundreds of graded predictions across the season. A favorite pegged near 56 percent that loses adds a small penalty to the running number, not a spike. The model's published MLB accuracy has been sitting around 68 percent on the season, and one game does not move that figure in any meaningful way.

Best call / worst miss

With one game on the slate, the best call and the worst miss are the same game, and there is no correct pick to celebrate. So this section is the miss.

The Phillies sent Aaron Nola to the mound and led their rotation matchup on paper, but the Mets got the better of it. Christian Scott threw 5 2/3 shutout innings for New York and the visitors launched three solo homers off Nola. Francisco Alvarez drove a 1-2 slider over the center-field fence in the third for the first, then homered again in the seventh, his 11th of the year, going back-to-back with Brett Baty. Nola had allowed just one run and three hits through six innings, but interim manager Don Mattingly sent him back out for the seventh at 90 pitches with a reliever ready, and the back-to-back shots flipped a 1-0 game into a 3-0 hole. The final was 4-1 New York.

The model liked the wrong side. The Phillies were the reasonable favorite, and the reasonable favorite ran into a sharp start from Scott and a catcher having a two-homer night. That is baseball, and it is exactly the kind of outcome a 56 percent favorite is expected to hand you a healthy share of the time.

What it means

Noise, not pattern. A one-pick day that lands on the wrong side of a coin-flip favorite tells you almost nothing about how the model is calibrated. If Verdexed graded this same matchup a hundred times at the same price, the favorite would be expected to lose it something like 44 times, and Thursday was one of those. The result is fully consistent with a model that is still hitting in the high-60s on the season.

The more useful context is that the full schedule returned on July 17, which means the next scorecard will grade a real slate with a proper spread of favorites, underdogs, and high-confidence flags. That is where the model's edge either shows up or it does not. A single graded game on getaway day from the break is a footnote, and it would be dishonest to spin it as anything more, in either direction. The model missed its one pick. The season number still reads around 68 percent. Both things are true.

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.