Yesterday's picks: how the model did (September 14)

Verdexed stated no sides on Monday, September 14, so the graded record for the day is zero for zero. Nothing was published to grade, nothing was scored, and the running ledger this column keeps sits exactly where Sunday left it. That is not a good day. A scorecard desk exists to grade model output, and on Monday there was none to grade.

The scorecard

Graded decisions: none, in either league that played.

**NFL.** Week 1 closed with a single game, Denver at Kansas City, and the Verdexed column published ahead of it carried no model probability, no edge, and no rated side. It said so in its own opening paragraph. Zero picks, zero correct, zero incorrect.

**MLB.** A Monday board in the last full week of the regular season, and again no model board. Zero picks, zero correct, zero incorrect.

**NBA and NHL.** Both remain in their offseasons, with nothing scheduled.

No accuracy percentage and no Brier score are cited here. The verdexed.com accuracy pages remain blocked from the environment this column is written in, and search returned the published methodology rather than the graded ledger underneath it. Rather than estimate a number, this column reports what it can confirm and leaves the rest blank.

The hand-kept ledger of stated sides, carried forward from Sunday, still reads 13-10 across 23 decisions. Monday added nothing to either column. Sunday's 1-3 remains the most recent day with a record attached to it.

What was readable Monday was the market half, which both columns did quote. Both quoted prices held.

**Kansas City minus 3 against Denver**, with the moneyline near minus 132 and a total quoted at 42.5. Result: Chiefs 31, Broncos 10. The favorite covered by 18 points, and 41 combined points went under the number.

**Chicago near minus 141 against Atlanta** at Wrigley Field. Result: Cubs 7, Braves 3. The home favorite held.

Those are market prices, not Verdexed's opinions, and they are reported here only because they are the one thing Monday's columns put a number on. A transparency desk should say how the numbers it printed turned out, whoever set them.

Best call / worst miss

There is no best call and no worst miss, because there was no call to make. That is the accurate answer and it should not be dressed up as anything else.

What can be graded is the reasoning the columns published without a price attached, and there the record is mixed.

**The NFL read does not survive contact.** The column described a market moving toward Kansas City without conviction, noted a spread of only three points, cited outside models that disagreed with each other, one making Denver a favorite by about three points and another putting Kansas City near a coin flip at 52 percent, and read the shape as a defense-first division game with a wider range of outcomes than the number implied. It paired Patrick Mahomes returning from a torn ACL and LCL with a ruled-out starting left tackle as reasons to doubt the home side in particular. The game was a 21-point rout in Kansas City's favor. Mahomes finished 15 of 27 for 184 yards with two touchdowns, an interception and a rushing score, and Kenneth Walker III ran for 173 yards. Bo Nix threw for 131 yards with a touchdown, an interception and a fumble. The column was right that three points understated the variance. It was wrong about which side the variance would land on, and it leaned its uncertainty in the direction that turned out not to matter.

**The baseball read held up.** The Dodgers and Reds column pointed at the Skubal and Lodolo rematch and at the low-scoring game the same pairing produced a week earlier. Tarik Skubal threw seven scoreless innings, allowed three hits and struck out nine, and Los Angeles won 4-1 to clinch a fourteenth consecutive playoff berth. Low-scoring was the read and low-scoring is what happened. It still was not a pick, and it does not go in the ledger.

What it means

One day with nothing to grade is not a model result. It is a publishing result, and that is the finding worth watching. Verdexed's methodology is explicit that the model reports win accuracy and a Brier score, and that a 65 percent call is meant to be a real 65 percent over a large sample. None of that has reached this column for days, which means readers are being asked to take calibration on description rather than on a number. That is a gap, and naming it is the job.

Against the longer run, the honest statement is narrow. Across the stated sides this column has been able to grade, the record is 13-10, and 23 decisions is far too small a sample to support a claim about edge in either direction. Sunday pulled it down. Monday left it alone. Neither day means much on its own, and nobody should extrapolate a trend from either.

The thing not to do is treat a blank day as a neutral one. A platform is accountable for what it publishes, and when the board is empty the correct entry is a zero and an explanation, not silence.

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.