New Jersey Devils vs New York Rangers Prediction: Rangers Edge

Monday night brings one of the NHL's most familiar rivalries back to the calendar, with the New Jersey Devils and the New York Rangers meeting at 23:00 UTC on 21 September 2026. Two metropolitan-area clubs, one result to settle, and no draw available as an outcome.

The headline from the TuringStats AI panel is a clear lean toward the visitors: across ten frontier models, the New York Rangers take 60% of the vote to the Devils' 40%. The most common predicted score is 0–2, and the panel grades this a strong-tier call, though the average confidence attached to it sits at a modest 18%.

Match context

The fixture is an NHL regular-season game between two long-standing rivals from the same metropolitan area, scheduled for Monday, 21 September 2026 at 23:00 UTC. The dossier provides no venue, no line-up information and no recent-form record for either side, so the panel's read rests entirely on the matchup itself rather than on a run of results.

What is notable is the shape of the split rather than its size. A 60–40 vote share is a majority, not a landslide, and the low average confidence of 18% tells the same story from a different angle: the models agree on the direction of the result, but they do not agree strongly on how comfortably it arrives. In a rivalry game with no form data to anchor the projections, that combination — a directional majority paired with thin conviction — is the defining feature of this forecast.

What the AI panel says

Ten models were run on this game. Six of them land on the New York Rangers, four on the New Jersey Devils, producing the 60% to 40% consensus. The average confidence across the panel is 18%, which places the call in the strong tier on vote share but well short of the certainty a higher confidence figure would imply. The single most common scoreline predicted by the models is 0–2 — a two-goal Rangers win featuring a Devils shutout.

The per-model picks reinforce how narrow the underlying disagreement is. The models splitting toward New York point to the Rangers' edge in controlling the run of play and their ability to keep the Devils off the scoreboard, which is why 0–2 recurs as the modal result. The models siding with New Jersey lean on the opposite reading: that in a rivalry fixture the gap between these two teams is small enough for the home side's structure and game management to close it. No model projects a high-scoring game, and none projects a blowout. That shared restraint — a low-event, tight-margin game — is the one thing the panel genuinely agrees on.

The panel's most common scoreline, 0–2, leans on New York's defensive control.

Key factors

Tempo and game state

The recurring 0–2 scoreline implies the models expect a low-tempo, low-event game rather than an open, chance-heavy one. When a panel converges on a two-goal margin that includes a zero, it is effectively describing a match decided by fine margins — a single breakthrough, a single mistake, or a single special-teams moment — rather than by sustained pressure at either end.

Defensive structure and the shutout lean

The zero on the Devils' side of the modal score is the most striking detail in the dossier. It reflects the models' view that New York's defensive structure is the most reliable element in this matchup, and that New Jersey's path to a result runs through breaking that structure down rather than through an exchange of chances. It is also why the confidence figure stays low: predicting a shutout is a demanding call, and the panel prices that difficulty in.

The rivalry effect

With no form data supplied, the models have little to separate these teams beyond the matchup itself. Rivalry fixtures between metropolitan-area clubs tend to compress the gap between them, which is consistent with a 60–40 split rather than a wider one. The four models backing New Jersey are, in effect, arguing that familiarity and the absence of a clear talent gap keep this game live for the full sixty minutes.

Where the panel is genuinely divided

The disagreement is not about whether this will be close — it is about who converts a close game. Six models see New York's control translating into the decisive goal and an empty-net or late second to seal it. Four see New Jersey surviving the pressure and finding the goal that flips the result. Both readings are internally consistent, which is precisely why the average confidence is only 18%.

New Jersey Devils vs New York Rangers prediction

The panel's verdict is a New York Rangers win, with a predicted score of 0–2 — matching the most common scoreline across the ten models. That scoreline is the modal result, not a mean, and it reflects the panel's shared expectation of a tight, low-scoring game settled by New York's defensive control.

Confidence should be read plainly. A 60–40 vote share is a real majority, and the panel classifies the call as strong tier, so the direction of the prediction is the panel's clear position. But an average confidence of 18% is low, and it matters: four of ten models back the Devils, and the most common score involves a shutout, which is inherently a demanding outcome to project. In practical terms, this is a forecast in which the winner is clear but the margin is not — a two-goal Rangers win is the single most likely result, yet a one-goal game either way sits comfortably within what the panel's split implies.

For the complete model-by-model output, including each pick and its reasoning, see the full model breakdown on TuringStats.

See every model's pick, confidence and reasoning for New Jersey Devils vs New York Rangers, or browse all upcoming Hockey predictions. Forecasts are informational only and are settled against the final result after the game.