One underdog a day, picked by a fixed rule off the model's raw number. Today it's Seattle at home against Toronto.
This isn't a parlay card and it isn't a gut lean. Sled Dogs runs one rule, every single day: among every underdog on the NHL slate — any team priced at positive American odds — we check which ones the model's own raw number, before any blending toward the market, rates at 50% or better to win outright. If more than one clears that bar, we take the one furthest above its break-even price. If nobody clears it, the column is dark. No forced plays, no filler picks to keep the streak alive.
Today, one team clears it. Toronto Maple Leafs visit the Seattle Kraken, and the dog in this one is Seattle.
Seattle is priced at +120 on the moneyline against Toronto, which means the market needs the Kraken to win 45.5% of the time just to break even on that number. The model's raw, unblended output puts Seattle's win probability at 56.1%. That's a 10.6 percentage point gap above break-even, and that gap is the entire case for the play. Nothing else. Expected value at this price works out to 23.4%, which is the number that gets a dog printed in this column rather than skipped.
To be clear about what that 56.1% means: Seattle still loses this game more often than it wins, even by the model's own math. The claim here was never that Seattle wins. It's that the market's price doesn't reflect what the raw model sees. That distinction matters more in this column than anywhere else on the site.
Here's the mechanical reason this column exists at all. The model's headline, published probability is blended roughly 55/45 toward the market price itself. That's useful for most purposes, but it has a side effect: when the market has already made a team an underdog, that blending pulls the model's number back toward the market's own skepticism, usually dragging it back under 50% before it ever gets a chance to disagree. Sled Dogs skips that step. It uses the model's raw, pre-blend read, the one that hasn't already been nudged toward agreeing with the price. That's the only way this column can ever find a team the market and the model see differently.
On this game, the factors feeding that raw number include away injuries hurting Toronto, home injuries hurting Seattle, away high-danger rate hurting Toronto, and home high-danger rate hurting Seattle. Every factor here works against somebody — there's no clean sweep in either direction, just a collection of drags that nets out to the number above.
The model raises its own flags on this one, and we're not burying them. Goalie starters aren't confirmed for this game. The model disagrees with the book on who the puck-line favorite should be. And there's no observed GSAx on either goalie to lean on. None of that invalidates the play, but all of it belongs on the table before anyone acts on a 56.1% number.
The column's record stands at 8-5 across 13 prior picks, +4.13 units at the prices taken. That includes wins like Anaheim at +105 and New Jersey at +100, and losses like Tampa Bay at +105 and Boston at +142. It's a small sample on one model's unblended output — not a demonstrated edge, and not financial advice. Treat today's Kraken play the same way: one number, one game, tracked openly either way it lands.