Yesterday's picks: how the model did (July 24)

Friday was a losing day at the top of Verdexed's MLB card. The single side the model was most confident in, Milwaukee at home against Colorado, lost outright, and a second published lean went down in a shutout. Of the three headline reads with a clearly stated direction, one landed and two did not.

This column grades the model in public whether the day was good or bad, and Friday was bad at the place that matters most.

The scorecard

Verdexed published five model reads on a full 15-game Friday. Three carried a clear direction and grade cleanly. Milwaukee at home against Colorado was the firmest win-probability side on the board, with the market pricing the Brewers near -243. It lost. Philadelphia at home against the Yankees was a modest lean built on a starting-pitcher edge. It lost. Los Angeles at the Mets, described as a narrower edge than the standings implied, won. One for three.

The other two published reads do not grade as wins or losses. Atlanta at Baltimore was flagged as a low-single-digit lean rather than a stand, with no side stated firmly enough to score. Cubs at Pirates was published with the explicit note that there was no meaningful edge and no side worth backing, which is the model declining a pick rather than making a bad one.

The most important item on the scorecard is the confidence tier, and it did not hold. Verdexed's own picks column had warned in advance that Milwaukee at -243 already had most of the edge priced out. That caveat was accurate, but it does not change the grade. The model's best side of the day lost.

A precise full-slate graded record for Friday, including the exact hit rate across all 15 games and the day's Brier score, was not retrievable from Verdexed's public accuracy surface at the time of writing, so no figure is claimed here. What can be verified is the published output, and on the published output Friday was a losing day.

Best call / worst miss

The best call was Los Angeles at the Mets. The Dodgers won 4-2, and the pick worked out for a reason the model had specifically flagged as a concern: Roki Sasaki was the projected starter carrying an ERA near 5.00, which is why the edge was written as narrow rather than wide. Sasaki then struck out nine over seven innings, set a career high with 21 whiffs, and allowed only a second-inning solo homer plus two other hits. Kyle Tucker homered and Los Angeles moved to 66-38. The side was right, though the projection's stated worry about the matchup turned out to be the very thing that decided the game in the model's favor.

The worst miss was Milwaukee. Colorado won 5-2, with Tomoyuki Sugano earning his 10th win and the Rockies scoring in the first, third and fourth innings to build a 4-1 lead. Solo home runs from Christian Yelich and Brice Turang were the only damage Milwaukee did. The model's firmest side of the day, a heavy home favorite against one of the league's weakest road clubs, was buried by the team it was supposed to beat. There is no reading of that other than a clear loss on the day's biggest number.

The Philadelphia miss was closer but no less a loss. The Yankees won 1-0 as Cam Schlittler struck out 12 and walked none over 7 1/3 scoreless innings. Jesus Luzardo, the pitcher the model's lean was built around, threw seven innings of one-run ball with nine strikeouts and lost anyway, with Paul Goldschmidt's solo homer the only run of the game. The picks column had flagged that the Yankees' Friday starter was unsettled between Will Warren and Schlittler, and Schlittler is who pitched. Flagging a risk in advance is good practice, but the pick still lost, and the flag does not convert it into anything else.

For the record on the game the model would not commit to: Atlanta beat Baltimore 7-6 in extra innings, scoring four runs in the 10th after the Orioles missed chances to close it out.

What it means

One losing Friday is one losing Friday. Thursday's scorecard had the model's top-rated side of that slate, Atlanta at 68.7%, holding in a 6-5 win over San Diego. Two consecutive days produced opposite results at the top of the ladder, which is what a thin-edge, market-anchored model actually looks like from close range.

The specific loss is also less unusual than it feels. A favorite priced near -243 is implied to lose roughly one time in four, so heavy chalk going down on a given night is inside the normal band rather than evidence of anything broken. What would matter is a sustained pattern of high-confidence picks underperforming their advertised rate across weeks, and a single Friday cannot show that. Verdexed's month-to-date accuracy figure was not retrievable at the time of writing, so no claim is made about the longer-run record.

The fair read on Friday: the top of the card failed, the one lean that landed did so partly for unexpected reasons, and the model correctly passed on a game where it saw nothing. Logged as a loss, not overreacted to.

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.