Wednesday night: a 4-3 NHL card, a split in college football, and a goalie move in Washington

The night in one line

Four wins, three losses, and +1.38u in the NHL. One win, one loss, and -0.09u in college football. No NFL games were played Wednesday. The best single result of the night was a puck line, the worst was a pair of totals that went over, and the most instructive part of the night happened before any puck dropped.

The puck line did the work

Colorado beat New Jersey 5-2, and the two Colorado results led the card: the moneyline won for +0.68u, and the puck line, Colorado -1.5, won for +1.64u, the largest single gain of the night. The puck line needed a win by two or more, and Colorado won by three.

Edmonton shut out Calgary 4-0, and the Under 6.5 there won for +0.93u as the strongest-rated lean of the night. Seattle edged Toronto 5-4, and the Seattle moneyline won for +1.14u at +114.

By market, the night split cleanly: moneyline 1-1 for -0.32u, totals 1-2 for -1.07u, puck line 1-0 for +1.64u. The puck line is the market that has carried the NHL card all season, and Wednesday was one more night where it did.

Washington: the goalie news moved the number

The Washington game is the one to study. The model's number moved overnight, when Logan Thompson replaced Charlie Lindgren as Washington's projected starter: Washington's win probability rose from about 59% to about 65%, and it sat at about 64% when the picks locked.

The locked leans went the other way. Philadelphia at +190 carried a small value edge, with the model putting Philadelphia's win chance at 35.8% against the market's 33.1%. The Under 6.5 carried a larger one, 61.8% against 55.1%. Washington won 5-2, seven goals in all, and both leans lost.

That is the difference between a projection and a lean. The projection said Washington was the likelier winner, and Washington won. The leans were price comparisons, spots where the model's number sat far enough from the market's number to clear the bar, and both went the wrong way on the night. A lean is not a prediction that the side wins. It is a statement that the price looked generous relative to the model's number. Prices can look generous and still lose, and on Wednesday both of them did.

The practical lesson is about timing: the model's number near the lock is the number that matters, because it holds the latest lineup and goalie news. A number from the morning is a different number.

College football: the flipped line won, the favorite did not

Two games, two different stories. In Murfreesboro, the spread flipped during the day, moving from Middle Tennessee -1.5 to Kennesaw State -1.5. The locked pick became Middle Tennessee +1.5, and Middle Tennessee covered by 14.5 points, winning by 13. The market moved toward Kennesaw State during the day. The model's number did not follow it all the way, and the gap between the two was the pick.

In Las Cruces, the model had Jacksonville State -3, and Jacksonville State lost by 18, missing the cover by 21 points. It was the night's clearest miss. The model had Jacksonville State rated well clear of the market, and the game went the other way by a wide margin. One game proves nothing on its own, and this one is recorded the same way as the wins: at the locked line, in the full record.

The two games together: 1-1, -0.09u.

The season picture

Through Wednesday, the NHL card stands at 156-143 and +13.90u: the puck line at 42-32 and +20.00u, the moneyline at 61-58 and -1.60u, totals at 53-53 and -4.50u. College football spreads stand at 79-96-3 and -24.18u, with totals tracked separately at 112-97-0. The NFL card, idle Wednesday, stands at 63-63-1 and +4.98u.

The trend read agreed with one of Wednesday's leans and disagreed with two, splitting 1-1 on the two it disagreed with. One night is not a sample, and the trends stay what they are: context around a pick, not a pick on their own.

How to read tonight's card

The same rules apply every night. Picks lock two hours before the start, and they are graded at the line they locked at, not the line at first publish, and not the closing line. The number to compare is the model's number against the market's number at the lock. Trends add context. And one night, good or bad, moves the season record by a little, because the season record is built from hundreds of picks, not one.

21+. If gambling stops being fun, call or text 1-800-GAMBLER.