It’s playoff time, and that means it’s time to talk odds of winning the first round matchups. Some recent research into why humans fail in their perceptions of probabilities poses an interesting answer to this question. It’s long been understood that we are actually pretty bad at probability, and we tend to turn complex problems into on/off answers and then complain when the weather forecast was “wrong”. We are also usually overly concerned with very small probability events, which is one reason why pretending a team that isn’t going to make the playoffs just might, maybe, possibly, if this and that and the other thing happens is a winning formula in sportswriting. See also: every prospect discussion ever.
What the new research suggests is that we’re not suffering from cognitive bias, we’re just doing the best our brains can do. What people are able to do is either pull back and take a big overview of probabilities with low accuracy, but covering a lot of potential results, or we can focus in on one set of decisions or choices or potential outcomes and perceive very fine differences. In a sense this is like being able to sort teams into broadly simple categories of good, bad and meh (since a team’s quality is really just its probability of beating an average team), and also see the finely detailed differences between any two teams, while having no real idea on a full ranking of all 31 of them.
When we bring our biases and emotions into the game as well, we really are bad at this, which is why gambling is an industry that succeeds very well for the house, but not the players.
We’re not gambling with money in a hockey playoff round, we’re gambling with our future feelings. Being a Blues fan is not a fun experience right now:
Moneypuck uses a season simulation model to produce individual game odds and playoff predictions, and you can read about that in full on the site. It is very distinct from gambling odds which are a past performance calculation influenced by how people are actually betting. We’ve already discovered that people are the last place you should go for good opinions on probability.
This is the second most lopsided round in Moneypuck’s full set of odds:
This is third:
Prediction: Oilers in 5 over Jets pic.twitter.com/mni0MggjQ3— MoneyPuck.com (@MoneyPuckdotcom) May 12, 2021
For reference, last year’s MP odds before the world went sideways had the Leafs likely playing Tampa Bay in round one, and likely losing, but Tampa’s edge was only 53.8 - 43.3 — it doesn’t add up to 100, because both teams has smaller chances of playing someone else at that point.
This season is weird, given the division arrangement, but the bottom line for round one is that the Leafs are overwhelming favourites to win it. So overwhelming that something like the last Boston series where the teams limped into a seventh game when there was a legitimate opportunity for the Leafs to close out the win in game six will be cause for a revolution, not a learning experience.
That Boston series in 2019 had a 51-49 weighting in favour of the Bruins, which made it a coinflip with the coin came down the wrong way.
This year’s round one is not coinflip territory. This series is something much more emotionally dangerous: a series the Leafs are expected to win easily. Even allowing for the randomness of hockey, this is a series the Leafs must win or there will be consequences beyond sadness.
That’s scary, so let’s check those odds with another source in case they’re wrong.
so my preview methodology:— Cole Anderson (@ice_cole_data) May 13, 2021
1) look at team season series by implied win probability
2) adjust for home-ice (worth about 2.9% this season)
3) take median team implied win % (this doesn't necessarily account for injuries/trades/timing/etc but close enough)
4) simulate series 10000x
This is broadly similar to MP’s method, but with different information feeding into the simulation.
the median EDM implied win probability vs WPG this season was 53.1% (that line was simply copied for their last 2 match-ups)— Cole Anderson (@ice_cole_data) May 13, 2021
over a 7-game series (with home-ice advantage) EDM wins 58.5% of the time https://t.co/8sQUN2Id5u pic.twitter.com/MBl7NDJWTK
The results are lower numbers for both Edmonton and Toronto, but still in scorching hot favourite territory for the Leafs. Anderson has Colorado at 77%, and if you hop into that Twitter thread, you can find all the others as well.
Dom at the Athletic has some similar work for subscribers, and hints at agreement with the above two models here:
Highest Stanley Cup odds per my model since 2010. Colorado is on an entirely different level this season. pic.twitter.com/B4jkzO662S— dom at the athletic (@domluszczyszyn) May 15, 2021
He’ll have a Leafs/Canadiens preview up at some point as well.
This is it, the hard truth we have to face up to: The Leafs are extremely good, better than ever before in my opinion — don’t tell me stories about that Leafs team that one time that actually had a second forward line, I don’t believe you. The easy days of having the Bruins as the ready excuse, of coinflips and game sevens are long gone.
The only acceptable outcome is a divisional win. No, that’s not a fair read of the probabilities because 70% is not a guarantee, and yes that’s emotional, not rational. But it’s also true.
Winning is the necessary outcome. Call me when you are Kings in the North, Leafs. That’s your minimum acceptable result.