Advanced stats vs. the eye test: Penguins are heavy favorites among Final Four

Even though the traditional eye test isn’t too enamoured by the Penguins, their overwhelming numbers are enough to make them the clear Stanley Cup favorite.

The final four is set: Pittsburgh, Tampa Bay, St. Louis and San Jose are the last teams left vying for the Stanley Cup. It’s an interesting and exciting group, but it’s also not an unexpected one if you paid attention to the numbers. Before every round, we here at THN previewed
three different sets of predictions: one based on stats, one based on the eye test, and a combination of the two. So far, the stats have come out on top and those four teams were pretty much exactly the predicted final four. Pittsburgh was the top team in the league. St. Louis was tops in the Central and fourth overall. Tampa Bay was tops in the Atlantic and seventh overall. And San Jose was second in the Pacific behind LA, but third overall in the league. Not bad for a bunch of numbers, let’s see how they do in the conference final. But first a quick recap of where the stats and eyes differed in opinion for round two, and some adjustments to the method.

WSH vs. PIT: This was the only series from round two where there was a divergence of opinion between the eyes and numbers and it turned into another win for the nerds. Both sides thought one team had a sizeable edge, but the series ended up being very even with just one goal separating the two over six games and three overtimes needed. That’s exactly how the combined method called it with a 51-49 split in Pittsburgh’s favour.
ADJUSTMENTS Once again, we need to do some fine tuning. In terms of probabilities, the numbers side has been much more accurate for the playoffs posting a log loss –
a measure of prediction accuracy – of 0.573 compared to 0.673 for the survey results (lower is better). The stats have “called” nine of 12 series correctly so far while the eye test has gotten seven of 11. They’ll now be weighted at an almost 60/40 split instead of 55/45 in favour of the numbers. It’s not just these two instances either. I kept track of the predictions of various other models as well as bookmakers, pundits, and public voting throughout the playoffs and the statistical models have been the most accurate of the four subgroups by far with an average log loss of 0.589. Bookies are next at 0.622, with the average pundit at 0.637 and the public at 0.693. It’s worth noting that the survey is the worst performing in the pundit subgroup so far, likely because the survey results, like the stats models, have not been updated during the playoffs. While the other experts can have their opinions shaped and moulded by what they see happen as the playoffs progress to make better judgements, the survey is a reflection of what some experts thought would happen before the playoffs began. That makes it a better comparative tool against the math models that don’t use any playoff data either. An even playing field. This is, of course, just one playoff year, a small sample size if you will, and therefore not a referendum of the entire stats vs. eye test debate. The numbers are kicking butt this year, sure, but that doesn’t diminish the value of the eye test. But it should be eye-opening for some of the doubters of the analytics movement. With that being said, let’s take a look at the four teams left vying for hockey’s ultimate prize, in order of likelihood.
PIT No team looks more impressive right now than the Penguins, especially after knocking out the Presidents Trophy champions in six games last round. THN writer/editor Matt Larkin is calling the Penguins this year’s
team of destiny and it’s hard to disagree with him. Pittsburgh has been on another level since Mike Sullivan was instilled as head coach and it’s all coming together perfectly in these playoffs. Separating Sidney Crosby, Evgeni Malkin, and Phil Kessel onto different lines has paid instant dividends as other teams simply don’t have the depth to handle what is essentially three first lines. That’s exactly what the numbers expected heading into the playoffs. Pittsburgh was easily the most impressive team down the stretch and have carried that play into the post-season. They were the favourites to win it before the post-season started, and that remains true today. Tampa Bay has looked impressive, but the Penguins shouldn’t have too much trouble against them and would be a favourite over both San Jose and St. Louis. The eyes don’t exactly agree. At least, they didn’t before the post-season started. The survey respondents saw a team with weak defence and goaltending. Defence and goaltending were certainly question marks before Round 1, but I don’t think that’s the case anymore. Matt Murray has been lights out and the defence by committee led by Kris Letang is doing just fine. The Pens should still narrowly edge out Tampa Bay, but they’d be an underdog in the Final no matter who comes out of the West according to the eyes. Even though the eyes aren’t too enamoured by the Penguins, their overwhelming numbers are enough to make them the clear favorite of the remaining four.
STL   Flip a coin. That’s how close these two teams are any way you slice it. By the numbers, the Sharks are 50.2 percent favorites. By the eyes, it’s the Blues at 50.4 percent. Combined? How about 50.03 percent. Yeah. That close. The numbers liked both teams a little bit more at the start of the playoffs, ranking them 3rd and 4th as opposed to 5th and 6th by the eye test. As it stands now, both teams have been very impressive throughout the post-season and especially in their deciding Game 7s. Huge victories in a do-or-die game show these two are past whatever issues plagued them in previous playoffs. They deserve to be where they are now. Unfortunately, one of them will have to go home while the other moves on. Good luck figuring out which one that’ll be at this point though. What we do know is that this will be a hard fought battle between two teams that have
killed the playoff narratives that have been attached to them for so long. Two former chokers, now one gets to play for the Cup.
TB The Bolts have probably been the most impressive team this post-season for how they handled things without two of their best players, but you have to consider who they’ve faced. Both the Red Wings and Islanders were unanimously considered one of the worst teams participating in the playoffs this season, and the Lightning were lucky enough to face both en route to the third round. Yes they’re 8-2, but those are important things to consider. Pittsburgh poses a much more difficult challenge for Tampa Bay, and without Steven Stamkos and/or Anton Stralman, it’s hard to see them making it close. That’s at least what the numbers think, giving Tampa Bay a 38 percent chance of advancing. And that’s without taking into account the massive injuries. The eye test thinks this will be a lot closer to a coin flip, with the Lightning still getting underdog treatment. And even if they do beat the Penguins, they likely won’t beat either of St. Louis or San Jose. The Lightning are a great team, but they’re probably the weakest left and playing the strongest team to boot. Round three could very well be the end of the line for Tampa Bay.

READ ALSO:  World junior championship 2017 preview: Finland