Yesterday's picks: how the model did (September 4)

Verdexed stated no MLB sides for Friday, so the graded line for September 4 is zero picks, zero correct, zero incorrect. That is thirty-one consecutive slates with nothing belonging to the model to score, and the honest first sentence of this column is that the model did not do anything yesterday, well or badly, because it did not say anything.

Baseball was the only place a Verdexed number could have been graded. The NBA and NHL are in their offseasons, and the NFL regular season does not open until Wednesday, September 9, when Seattle hosts New England at Lumen Field. One league played, and the model was silent on it.

The scorecard

Zero stated directions, zero correct, zero incorrect for September 4.

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

The reason is the same one this page has been printing for a month. The verdexed.com prediction, edge, and accuracy surfaces are unreachable from this publishing environment, and per-game win probabilities, confidence tier assignments, and Brier values were not retrievable through search either. So there is no answer here to the question this column exists to answer on a normal day, which is whether the 65 percent and 70 percent confidence bands held. No trailing accuracy figure and no Brier value are claimed, because none was retrievable. The methodology page verdexed.com publishes still describes an MLB win accuracy near 68 percent with a scored Brier alongside it. That is the site's claim, not a number this page has verified in a month, and readers should treat it as unverified here.

What did get recorded Friday morning was the market. The picks column quoted three prices across books before first pitch, and those three are gradeable even though none of them is a model side. Atlanta was near minus 152 at Philadelphia, Tampa Bay near minus 122 at Texas, and the Mets near minus 128 against San Francisco. All three favorites won. Quoted chalk went 3-0.

That is a market ledger, not a Verdexed ledger, and three prices is a sample small enough that 3-0 and 0-3 are both ordinary outcomes.

Best call / worst miss

There is no best call and no worst miss belonging to the model, because no side was stated. Naming one now would be inventing a record.

The firmest number on the board did land. Chris Sale outpitched Cristopher Sanchez at Citizens Bank Park and the Braves beat the Phillies **5-2**, allowing two runs over six innings with seven strikeouts. Ronald Acuna had three hits and Mauricio Dubon homered, while Alec Bohm drove in both Philadelphia runs. Atlanta's National League East lead grew to five games, which was the actual stake of the game the market had priced as its strongest opinion.

The quoted favorite that came closest to failing was Tampa Bay. The Rays beat the Rangers **7-6** after Elias Diaz hit a seventh-inning grand slam that cut a comfortable margin to one run. Nick Martinez went six and a third for the win, and Kumar Rocker was charged with six runs across five innings. A minus 122 price that survives a grand slam is a reminder that a favorite holding says nothing about how the probability was earned.

The two results a starter-aware model would most want credit for went unpriced here. Ranger Suarez threw seven scoreless innings and the Red Sox beat the Orioles **1-0** at Camden Yards, with Mickey Gasper's home run the only run of the game. In Seattle, the Athletics hit four home runs in the third inning off Logan Gilbert, reported as a Seattle franchise record for one inning, and won **7-6** in a game whose posted total of 7.5 was cleared by more than five runs. Milwaukee beat Cincinnati **10-7**, and the Mets beat the Giants **10-6** behind five home runs and a third straight win for Nolan McLean.

Six confirmed finals, one shutout, two one-run games, and no Verdexed number on any of them.

What it means

Pattern, not noise, and the pattern is not about baseball. A single unreadable day is a retrieval problem. Thirty-one in a row is an infrastructure problem, and the correct read of Friday is not that the model had a quiet day but that it had no day at all.

The 3-0 on quoted prices should not be dressed up as a win. It is the market's record, not the model's, and this desk did not stake anything on it. Reporting it as if it were a good day for Verdexed would be exactly the sort of borrowed credit an accountability column exists to refuse.

The ledger that does exist stays at 12-7 from a stretch that ended August 4. Against a market this efficient, 19 decisions proves nothing, and it will keep proving nothing until the model is stating sides again and those sides are scored on schedule. The fix remains plumbing rather than analysis: the prediction, edge, and accuracy surfaces need to be readable from where these columns are written. Until then, the honest daily entry is 0-0, and this page will keep printing 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.