By Ashkon Baharlou

Three games on night two of the season, Wednesday night, and the Whiteboard's read is blunt: its surest pick lost 7-0, and both favorites needed the other team's goalie to win.

Pittsburgh 7, Philadelphia 0: Filip Hallander had a goal and two assists. Expected goals said Pittsburgh, 4.65 to 2.14. Toronto 2, NY Islanders 1: Easton Cowan had a goal and an assist. Expected goals said the Islanders, 3.18 to 2.90. Colorado 8, Los Angeles 4: Nathan MacKinnon and Artturi Lehkonen scored twice each. Expected goals said Los Angeles, 3.38 to 3.03.

A quick note on the numbers, since they carry this article. Expected goals (xG) is the value of every chance. Individual expected goals (ixG) is a player's own number from the shots he took. Goals saved above expected (GSAx) is what a goalie stopped beyond the chances he faced — a positive number means he saved more than expected, a negative one means fewer. And on-ice share is a player's slice of 5-on-5 expected goals while he was out there. Power play covers 5-on-4, 5-on-3, and 4-on-3; penalty kill covers the reverse. All of it is the Whiteboard's tracking of the three games, built on the NHL's play-by-play.

Pittsburgh 7, Philadelphia 0

This was Pittsburgh's game from start to finish: 4.65 expected goals to 2.14 at all strengths, 3.84 to 1.04 at 5-on-5, a 79 percent share of the 5-on-5 chances, with 10 high-danger chances to 5 and 38 shots to 19. In plain terms: Pittsburgh created more than twice the quality of chances, doubled Philadelphia in shots, and owned nearly four of every five even-strength chances. The pick was Philadelphia at 80.4 percent — four chances in five to win — built on Philadelphia's skaters rating 64.3 to Pittsburgh's 58.9 and on Arturs Silovs grading 12.3, the lowest of the night's six starters. Both reads missed. Every Pittsburgh skater finished on the right side of 5-on-5 play at on-ice shares of 64 percent or better, and Silovs stopped all 19 shots on 2.14 expected goals faced, plus-2.14 goals saved above expected — about two goals better than those chances normally produce. Dan Vladar allowed seven goals on 4.65 expected, minus-2.35 goals saved above expected, about two and a half goals worse.

Pittsburgh's depth carried it. Filip Hallander had a goal and two assists on 0.81 individual expected goals, the most of any skater in the game, with an 85 percent on-ice share in 8.5 minutes at 5-on-5. That 85 percent means that while Hallander was on the ice, Pittsburgh owned 85 percent of the quality chances. Hendrix Lapierre, on that line, had a goal and an assist at a 95 percent on-ice share, on the ice for 1.35 expected goals for and 0.07 against. The Samuel Girard and Kaedan Korczak pair posted 85 and 86 percent shares, with two assists apiece. Nick Robertson scored twice, one of them shorthanded, on 0.36 individual expected goals. Blake Lizotte generated 0.57 individual expected goals on four shots without a point.

Philadelphia had nothing at 5-on-5. Porter Martone was on the ice for 0.09 expected goals for and 1.61 against, a 5 percent share, without a point. Trevor Zegras and Owen Tippett were each at 9 percent, and Travis Sanheim, the top-rated skater on either side, at 15 percent, on the ice for 1.91 expected goals against, the most in the game. Noel Acciari was the only Philadelphia skater above 50 percent, at 56. Nobody on Philadelphia had a point.

Special teams: Pittsburgh went 2 for 4 on the power play on 0.60 expected goals in about 6.9 minutes, with no high-danger shots, so both goals were finishing — two goals on chances worth just over half a goal. Philadelphia went 0 for 3 on 0.99 expected goals in about 7.0 minutes; Tippett's 0.27 individual expected goals led its power play. Robertson's shorthanded goal came on Pittsburgh's 0.20 shorthanded expected goals.

The surest pick of the night was its worst. Pittsburgh controlled play at every strength and got perfect goaltending from the night's lowest-graded starter.

Toronto 2, NY Islanders 1

The Whiteboard barely leaned Toronto at 61.0 percent, with Ilya Sorokin named as the risk — the Islanders goalie who could steal the game. The Islanders had the better of the chances at all strengths, 3.18 expected goals to 2.90, and 5-on-5 was even at 1.91 to 1.84, with six high-danger chances to five for Toronto. In other words: dead even on chance quality, with the Islanders holding a small edge overall. Sorokin was good, 24 saves on 26 shots, 2.90 expected goals faced, plus-0.90 goals saved above expected, about one goal better than expected. Anthony Stolarz was better: 26 saves on 27 shots, 3.18 expected goals faced, plus-2.18 goals saved above expected, about two goals better — and that difference decided it.

Toronto's top and third lines did the work at 5-on-5. Jack Roslovic was on the ice for an 82 percent share, Auston Matthews 77 percent, and Kirill Marchenko 65 percent; Roslovic had an assist, Matthews and Marchenko no points. The third line was in on both Toronto goals: Colton Sissons scored at even strength on 0.55 individual expected goals at a 69 percent share, Nick Paul assisted at 67 percent, and Easton Cowan, at 61 percent, assisted on Sissons's goal and scored Toronto's other on the power play. That is the same third line rated at 36 chemistry, which outplayed its rating on opening night too.

Toronto's second line was buried. William Nylander was on the ice for 0.13 expected goals for and 1.03 against, an 11 percent share, John Tavares 15 percent, and Gavin McKenna 23 percent; none of the three had a point. Chris Tanev was on the ice for 1.06 expected goals against at a 34 percent share. For the Islanders, Kyle Palmieri generated 0.59 individual expected goals, Anthony Duclair 0.45, and Victor Eklund 0.43, all without a point. Emil Heineman scored the Islanders' goal shorthanded. Matthew Schaefer played 19.2 minutes at 5-on-5 at a 49 percent share.

Special teams: Toronto went 1 for 4 on 0.83 expected goals in about 8.4 minutes. The Islanders went 0 for 4 on 0.65 expected goals in about 9.1 minutes, and managed only two power-play shots on goal. Heineman's shorthanded goal came on 0.24 of his own expected goals while killing the penalty.

The Whiteboard got the winner, but this was a goaltending game. The chances were even; Stolarz was better than Sorokin.

Colorado 8, Los Angeles 4

The pick, Colorado at 77.7 percent, came in, but not the way it was drawn up. Los Angeles had the better chances, 3.38 expected goals to 3.03 at all strengths and 2.23 to 2.03 at 5-on-5. The case for Colorado was its defense and Scott Wedgewood in goal; Mackenzie Blackwood started instead and finished at minus-0.62 goals saved above expected, allowing four goals on 3.38 expected — nearly even with expectations. Darcy Kuemper allowed eight on 3.03 expected, minus-4.97 goals saved above expected, the worst goaltending line of the night — about five goals worse than those chances normally produce.

Colorado won it on the power play: 4 for 5, on 1.00 expected goals in about 7.2 minutes, including about half a minute with a two-man advantage. Four goals on chances worth about one — the finishing, not the chance quality, decided this one. Artturi Lehkonen scored twice on the power play on 0.63 individual expected goals there. Nathan MacKinnon scored twice, once on the power play, on 8 shots and 0.65 individual expected goals, 0.14 of it on the power play. Brock Nelson had a goal and two assists. Cale Makar, Nazem Kadri, Devon Toews, and Jaden Schwartz each had two assists. At 5-on-5, Colorado's best numbers came from Sam Malinski, a 68 percent share, and Nicolas Roy, who scored at 66 percent.

The third line rated the night's weakest held true. T.J. Hughes was on the ice for 0.02 expected goals for and 0.75 against, a 3 percent share, and Fedor Svechkov was at 30 percent; neither had a point. Noah Juulsen was at 16 percent.

For Los Angeles, the defense was split. Brandt Clarke posted an 82 percent share, on the ice for 1.03 expected goals for and 0.23 against, and Joel Edmundson 73 percent, while Drew Doughty, in 16.5 minutes at 5-on-5, was at 25 percent with an assist. Adrian Kempe generated 0.54 individual expected goals, most of them on the power play, without a point. Andre Lee posted a 79 percent share in limited minutes. Artemi Panarin, Brian Dumoulin, Trevor Moore, and Joel Armia scored; Moore's and Armia's goals were shorthanded, on 0.11 Los Angeles shorthanded expected goals.

Special teams: besides Colorado's 4 for 5, Los Angeles went 0 for 4 on 0.98 expected goals in about 9.0 minutes, with Panarin's unit on the ice for 0.88 of them.

The winner was right for the wrong reasons. Los Angeles carried the play; Colorado's power play and Kuemper's night decided it.

The standouts

  • Arturs Silovs, Pittsburgh: a 19-save shutout on 2.14 expected goals faced, from the night's lowest-graded starter.
  • Darcy Kuemper, Los Angeles: eight goals allowed on 3.03 expected, minus-4.97 goals saved above expected, the worst goaltending line of the night.
  • Colorado's power play: 4 for 5 on 1.00 expected goals; Los Angeles answered with two shorthanded goals on 0.11.
  • Filip Hallander, Pittsburgh: a goal, two assists, and 0.81 individual expected goals in 8.5 minutes at 5-on-5, at an 85 percent on-ice share.
  • Toronto's second line: Nylander, Tavares, and McKenna at on-ice shares of 11, 15, and 23 percent, while the third line, rated weakest, was in on both Toronto goals.

What the crowd sees vs. what the model sees

The crowd sees a 7-0 blowout and two favorites winning, and reads a normal night. The Whiteboard's counter-read is that none of the three games went the way it expected. Its surest pick of the night, Philadelphia at 80.4 percent, lost 7-0 in a game Pittsburgh controlled at every strength. Toronto and Colorado were right on the result and wrong on the process: Toronto won a goaltending game with the chances even, and Colorado won on a 4-for-5 power play while Los Angeles carried 5-on-5.

None of the three games went the way the Whiteboard expected, not even the two it picked correctly.

The fair concession, and it matters: the ratings are still built on last season's data. Nothing from these games is in them yet. Two nights will tell us more than one did.

The verdict

The surest pick lost 7-0, and the two winners won the wrong way. Pittsburgh controlled everything against the 80.4 percent favorite and got a shutout from the night's lowest-graded starter. Toronto and Colorado took the results while the chances went the other way: even chances in Toronto decided by Stolarz, better chances in Los Angeles buried by Kuemper's minus-4.97 goals saved above expected and Colorado's 4-for-5 power play.

The real risk, named plainly: it is two nights. None of these games are in the ratings yet, and one blowout does not make Pittsburgh the better team. The chances tell you what happened while the scoreboard tells you who won, and on night two the Whiteboard misread all three.

Want to check the work? Open the Whiteboard, the free line tool I built, and pull up any of these games yourself. Build tonight's lineups while you are there and watch the ratings move.

All expected-goals numbers from the Whiteboard's tracking of the three games, built on NHL play-by-play. Scores, scorers, and goalie lines via the NHL's official data.

Follow @lastchangehockey on Instagram for more Whiteboard breakdowns.