One underdog a day, chosen by a fixed rule off the model's raw number. Today it's New Jersey at Dallas.
This isn't a card. It's not a lean, and it's not a parlay leg buried in a bigger post. Sled Dogs runs one rule every single day: among every game on the slate, find the team the market prices as an underdog — positive American odds — that the model's own raw number, before any blending toward the market, rates at 50% or better to win outright. Among any that qualify, we take the one furthest above its price's break-even point. If nothing clears that bar, the column goes dark. No forced plays, no backup dog.
Today, one team cleared it: New Jersey Devils moneyline at +100, on the road at Dallas.
Dallas is favored. New Jersey is priced at +100, which needs a 50.0% win rate to break even. The model's raw, unblended number puts the Devils at 60.0% to win this game outright — a full 10 points above that break-even line. That gap is the entire thesis. Nothing else.
Why raw and not the model's published, market-blended number? Because the public-facing probability gets pulled roughly 55/45 toward the market price, and on a team the market has already made an underdog, that blending drags the number back under 50% almost every time. It sands the disagreement right off. Sled Dogs skips that step and uses the model's unblended read, which is the only way this column can ever disagree with the market at all. Expected value at +100 with a 60.0% raw read works out to 19.9%.
None of that means New Jersey wins tonight. An underdog is priced as an underdog because the market thinks it loses more often than it wins, and a 60% number still loses plenty. The claim here isn't that the Devils win — it's that +100 undervalues how often the model's raw number says they will.
The factors aren't unanimous, and we're not pretending they are. Home injuries at -0.42 hurts Dallas. Home 5v5 expected-goals-for versus Dallas's expected-goals-against sits at -0.136, also hurting Dallas. Home high-danger rate at -0.132 hurts Dallas as well. Working the other direction, home-ice advantage at +0.045 helps Dallas — a small, standard bump that doesn't erase the rest.
The model raises its own caveats here, and they matter: goalie starters aren't confirmed for this one, the model disagrees with the book on who it likes on the puck line, and there's no observed goals-saved-above-expected data for either goaltender to lean on. Treat the raw number as a read with real gaps in it, not a lock.
Sled Dogs is 7-5 across 12 picks so far, +3.13 units at the prices taken. That's the whole sample — tracked separately from the model's own headline record and separate from any other picks column on this site. Recent results: Calgary Flames +130 won on 10/26, Anaheim Ducks +102 won on 10/24, Tampa Bay Lightning +105 lost on 10/23, Boston Bruins +142 lost on 10/20, New York Islanders +110 won on 10/17.
Twelve picks is nothing close to proof of anything. It's one model's output on a small sample, and this is analysis of that output — not a recommendation to bet your rent on the Devils tonight.