Yesterday's picks: how the model did (August 18)

There is no graded line for August 18. Verdexed's column for Tuesday's MLB board stated no sides, claimed no edges and assigned no confidence tiers, because the model's per-game probabilities were not retrievable when it published. The honest entry is zero picks, zero correct, zero incorrect, and that makes fourteen consecutive slates with nothing for this desk to grade. Baseball was again the whole graded universe: the NFL preseason had no Tuesday games, with Week 2 not opening until August 20, and the NBA and NHL remain in their offseasons. What can be counted is what the column actually put on the page, which was the market's ranking rather than the model's. Both games carrying a confirmed two-way price went to the favorite.

The scorecard

Zero stated directions, zero correct, zero incorrect for August 18.

The stated-side ledger does not move. It stands at 12-7 across 19 decisions, unchanged since August 4.

No trailing accuracy figure and no Brier value are claimed here, because none was retrievable. Verdexed's public accuracy surface could not be reached at publication, and no per-tier number for August 18 was available anywhere else this desk could source. That is fourteen slates without a quotable calibration figure, and no high-confidence tier to report on, because no confidence tier was published to begin with. A process that is not published is not a public record, and this column will keep saying so rather than gesture at a number it did not read.

The market is a separate instrument and earns the model no credit in either direction. It is the only thing Tuesday's column quoted, so it is the only thing that can be scored.

Los Angeles, the firmest price on the board at roughly 71 percent once the house margin comes off, beat Colorado **7-6** at Coors Field. San Diego, quoted at -112 against a -104 Mets side and netting out near 51 percent, beat New York **5-2** at Citi Field. That is 2-0 for the quoted prices, on a sample of two.

Three other games on the column's list carried no confirmable price and are not scored. Milwaukee beat Seattle **22-0**, a result Bleacher Report reported as the largest shutout margin in the majors since 1888, with 22 hits, four home runs and 21 runs across the final four innings. Minnesota beat Atlanta **4-1** at Target Field. Texas beat Washington **5-0**, behind seven innings and four hits from Cal Quantrill.

Best call / worst miss

There is no best call and no worst miss, because no side was stated. Assigning one now would be inventing a record after the fact.

The strongest quoted price did hold, and it held by exactly the margin the column warned about. Coors Field was named in Tuesday's edition as the standing risk to a 71 percent Dodgers number, and Los Angeles won by one run. The tightest quoted game, San Diego at New York at a rounding error above a coin flip, also went to the nominal favorite, with Fernando Tatis Jr. homering twice and accounting for three of the five San Diego runs.

The column's own worst miss on Tuesday was an input. It listed Bailey Ober as Minnesota's starter against Atlanta. Ober started Monday. Zebby Matthews started Tuesday and went five and two-thirds innings, allowing one run on five hits, while Josh Bell's two-run homer in the first and Ryan Jeffers' two-run double in the eighth carried the Twins to a **4-1** win. Drake Baldwin's sixth-inning home run was Atlanta's only run. A desk that grades a model's inputs in public has to grade its own, and the listed starter was wrong.

The second thing worth flagging against the column's own text: it quoted an independent model, not Verdexed's, giving Milwaukee roughly a 53 percent chance against Seattle, and offered it as evidence that a first-place team does not automatically draw a firm price. That game ended 22-0.

What it means

Two quoted favorites holding is not evidence of anything, and neither was Monday's 2-1 or Sunday's 3-0. Two games is a sample that supports no read at all, and a one-run Dodgers win at altitude is the ordinary shape of Coors, not a signal about pricing.

The 22-0 is the more instructive result. A model that priced Milwaukee near 53 percent was not wrong by 22 runs; probability estimates describe distributions, and blowouts live inside the same distribution as one-run games. That is why accuracy and Brier are measured across hundreds of predictions rather than off a scoreboard, and why this column will not celebrate a 2-0 Tuesday.

The number that would mean something is the one that still cannot be published. The stated-side ledger has been frozen at 12-7 since August 4, and a frozen record is not a good record, it is an absent one. No trailing thirty-day accuracy figure and no Brier trend can be cited here for the fourteenth straight day. Readers who want the probabilities, confidence tiers and graded accuracy these recaps are built to report should go to Verdexed's prediction, edge and accuracy pages directly.

This desk resumes grading the model the day the model's output is retrievable again, and it will report that day's result whichever way it falls.

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.