Washington generated superior expected goals and more high-danger chances but lost 3-2 in overtime — M. Blackwood's net turned away every chance that mattered.

The Capitals held 48% of shot attempts but generated 3.07 expected goals, a quality advantage. Colorado generated 2.88 xG on 24 shots. Washington's danger per attempt was higher: 0.073 xG per shot versus Colorado's 0.052. Yet on high-danger chances, Washington managed 0-for-5. Colorado converted 2-for-4.

M. Blackwood stopped 1.07 goals above expected, including a perfect 5-for-5 save rate on high-danger shots. L. Thompson saw more danger per attempt but saved only 21 of 24.

The Capitals' efficiency off the rush (2 goals on 26 chances) matched Colorado's (3 goals on 25 chances). The advantage dissolved in the scoring areas. Washington generated 5 high-danger chances to Colorado's 4. B. Nelson tipped in at 01:22 of overtime — the game's final word.

This was a variance clinic. Washington's underlying metrics suggested they'd escape with a point. The Capitals had the shot quality and quantity to demand it. Instead, they encountered an opponent that didn't need many chances, plus a goaltender who stopped everything dangerous.

The model projects a 2.8-2.6 game with the Avalanche at 54% win probability—essentially a pick'em.

Colorado owns the underlying play. At 5v5, the Avalanche generate 57.0% xG and drive 57.0% Corsi, controlling both territory and shot quality. Washington's 50.0% xG reflects a team playing catchup. That gap would usually settle a matchup decisively. Not this one, because both teams are running above expected luck.

Colorado's 1.021 PDO and Washington's 1.018 PDO mean both are converting at unsustainable rates and allowing fewer goals than their defense should permit. In a true coin flip, the expectation is both regress equally. When Colorado's superior play meets equal regression for both teams, the advantage cancels.

The scorers demonstrate this distribution. Artturi Lehkonen carries Colorado at 1.11 xG/60 with 18 goals and 14.9% shooting—he's earned his output. Ross Colton, at 1.10 xG/60, has only 7 goals on 4.9% shooting, leaving room for variance in either direction. Alex Ovechkin leads Washington at 19 goals, but his 0.97 xG/60 and 13.8% shooting rate show he's running ahead of his opportunities. Anthony Beauvillier, at 14 goals on 10.4% shooting, similarly exceeds his 0.89 xG/60.

M. Blackwood's back-to-back status is the sole lever. If Washington starts a backup or Blackwood plays tired, Colorado gets an edge. If Blackwood is fresh, the scales balance.

The verdict: when unsustainably hot teams both cool down, the one with superior underlying play wins. When luck persists, the lucky team does. This is a matchup where a coin flip is the most honest forecast.