The sharpest fact out of Wednesday's slate: the market closed Colorado Avalanche moneyline at -187 against Winnipeg, a number that says 65 percent win probability, and the Avalanche lost 2-3. That's not a puck-luck bounce. That's the board pricing in a gap that didn't show up on the ice.

I had that one too. Took -178, closed -187, beat the close by a full tick. Didn't matter. The game still went the other way. That's the first receipt of the night and it's not a flattering one: right side on the number, wrong side on the board, and the loss counts the same either way. Call it a defensible miss, not a bad read, the market and I were both leaning the same direction and the market was wrong with me.

NHL line movement today: where the close missed and where it didn't

The rest of the day cleaned it up. Three winners against that one loss, and on every pick that carried a closing number, I got a better price than the market settled at. Four for four on CLV. That's the stat that actually matters over a season, win or lose on any single night, because beating the close consistently is the tell that the number was sharp going in, not lucky coming out.

PickTookClosedResult
Colorado Avalanche ML-178-187Lost, 2-3
Washington Capitals ML-144-155Won, 5-3
Florida International ML-223-240Won
Edmonton Oilers ML-122-130Won, 5-2

Washington closing at -155 off a -144 entry and covering it at home against Pittsburgh is the clean version of the night: the market moved toward my number, and the Capitals backed it up with a two-goal win. Edmonton did the same thing on the road in Anaheim, book-beating price and a three-goal final. The Florida International line isn't hockey, it rode along on the same card and the number held the same way: took the better price, market closed shorter, ticket cashed.

So the night nets out at 3-1, +0.96 units on a flat stake, with the only loss coming on the game where the board and I both got embarrassed by a 7-seed mentality from Winnipeg that neither of us priced correctly.

One stray number worth flagging for color, not for the record: the single biggest pool of money anywhere on the slate was 1,809,986 in prediction-market volume sitting on Guardians at White Sox, a reminder that baseball's October money still dwarfs an NHL Wednesday even in a sport that's heating up fast.

What tonight's board should learn from it

A 65 percent favorite losing isn't rare, it's supposed to happen roughly a third of the time by definition, that's the whole math of a -187 price. The mistake would be treating one bad beat as a signal the model is broken. It isn't. The CLV says the model's still finding value before the market catches up. The record says the hit rate is fine. What it says for tonight is simpler: price the favorite, respect the variance, and don't let one ugly final change a number that was built right. The math says pass on overreacting to this one.

Nothing deleted, nothing cherry-picked. The Avalanche loss is on the board right next to the three wins, same as it'll be tomorrow if the next one goes bad too.