Yesterday's picks: how the model did (MLB, September 20)

Verdexed's graded line for Sunday, September 20, is zero picks, zero correct, zero incorrect. Baseball played a full Sunday card that clinched a division, clinched a postseason berth and closed out three home schedules, and not one game on it was scored by this desk, because the model's per-game probabilities were not published. By this column's own running count, that is the forty-fifth consecutive slate opening with the same number, and it is a bad result rather than a neutral one.

The scorecard

Picks made: zero. Correct: zero. Incorrect: zero. Accuracy on the day is undefined, because the denominator is zero.

Verdexed's best bets column for September 20 opened by stating that per-game win probabilities, confidence tiers and edge figures were not retrievable, that no edge was being claimed, and that no probability it printed belonged to the model. It listed Sunday's games for their stakes and their prices. There is nothing in it to grade as a pick.

No accuracy percentage and no Brier score is cited here. The verdexed.com prediction, edge and accuracy pages remain unreachable from the network this column publishes behind, and a public search returns the stated methodology, which describes scoring every logged prediction on win accuracy and on Brier score, rather than the record underneath it. That is a process, not a number this column could read. The 65 percent and 70 percent confidence tiers went unpopulated for a forty-fifth straight day.

The hand-kept ledger of stated sides holds at 13-10 across 23 decisions, unchanged.

One game on Sunday's board carried a quoted price. It lost.

Best call / worst miss

Nothing was picked, so there is no sharpest call and no worst miss of the model's. What follows grades the market read the column printed.

**The price that failed, in the day's biggest game.** The column quoted Houston at minus 138 against Atlanta, about 56 percent with the book's margin removed, and called it a lean rather than a conviction. Result at Daikin Park: **Braves 4, Astros 2**. Mike Yastrzemski hit a three-run homer in a four-run sixth, and Atlanta clinched the National League East for the seventh time in nine years. Hunter Brown, the projected Houston starter the column named, scattered three hits and struck out 10 over five scoreless innings, and Houston lost anyway. That is the ordinary reminder that a starting pitcher is a feature, not an outcome. The loss carried further than the standings line suggests: it clinched a postseason berth for Boston, and it came in a race the column had flagged as genuinely unresolved.

**The best call was a caveat rather than a forecast.** The column named Patrick Sandoval and Griffin Jax as projected starters for Red Sox at Rays and told readers explicitly that the Tampa Bay end could not be corroborated and should be treated as unconfirmed. Both started. Jax threw five shutout innings, Sandoval gave up four runs during a 40-pitch first, and **Rays 5, Red Sox 1** followed the shape of that pitching gap almost exactly, with Jonny DeLuca and Chandler Simpson driving in a pair apiece in the first and Victor Mesa Jr. homering. David Peterson, listed as the Cubs' projected starter under the same caveat, went seven innings and allowed one run with seven strikeouts in **Cubs 9, Reds 1**, a win that reduced Chicago's magic number to one and included Pete Crow-Armstrong's 45th home run. Being right about an unconfirmed lineup card is a small thing. On this page it is the only thing resembling a correct forecast.

At Citi Field, **Phillies 7, Mets 2** closed New York's home schedule at 36-45. The column had flagged that game as the classic spot where a model and a market diverge, because motivation is not a feature either one measures well. Without a model number, it stayed a description of a price rather than a disagreement with one.

What it means

Sunday moved the model's record by nothing, because Sunday never entered it. Going 0 for 1 on quoted prices is not a result worth interpreting either; one moneyline is noise, and the Astros losing a game their starter dominated is the most ordinary thing in baseball.

The pattern that matters is the one this column has now reported for six straight weeks. Verdexed moved per-game predictions behind a subscription at the end of July, and the public ledger has been unreadable since. The hand-kept 13-10 across 23 stated sides is the only record on this page that was ever graded, it is a small sample, and it is stale. With nine days left before the Wild Card Series open on September 29, the games are getting more valuable to grade and the board is no more readable than it was in August. That is worth saying plainly rather than filling the space with prices.

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 of legal betting 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.