Now that the regular season is winding down and the NHL playoffs are almost upon us, many fans are looking back at the thrilling victories and crushing losses of recent postseasons. Inevitably, these moments are the ones that define sports for fans – the wins that still bring a smile to your face years after the fact, the losses that take you days (if not longer) to get over. But part of what makes them so exciting is the element of randomness that determines who raises the big trophy at season’s end. In the regular season, the hockey gods have time to giveth and taketh away pretty equally across teams, and success is largely determined by the quality of the roster and the coaching. Yet the better team doesn’t always come out ahead in a seven-game series. A handful of unlucky bounces, a cold streak from a key scorer, or a hot opposing goaltender can put a superior team into a hole they can’t recover from. Using the model I’ve developed for the upcoming postseason, what if we took some historical data to figure out which upsets from seasons past were the most improbable?
As a quick refresher, my playoff model uses a modified measure of GF% (mGF%) to estimate team quality: 5-on-5 shots for and shots against are combined with even-strength Sh% and Sv% (regressed to team-specific three-year averages) to measure GF and GA at evens. Power-play goals for and against are also included, using the league-average rates of conversion for both. This measure is then combined with estimates of home-ice advantage effects to calculate probabilities of series victories. Calculating mGF% for every playoff team going back to 2008-09 and estimating these probabilities yielded interesting results. Specifically:
- This method of predicting series was accurate 69.3% of the time, though small-sample-size warnings apply. It was just 5-7 in series that were effectively coin flips (i.e., the favored team had a win probability of less than 52%), so the long-run performance may be a bit better.
- Had they been able to hold onto their 4-1 lead in Game 7 in Boston (mGF% 55.2%), last year’s Maple Leafs (47.6%) would be able to claim the most unlikely playoff upset in the past five years. The Bruins entered that series with an estimated 69.7% win probability.
- The model would not have been surprised by the quality of the Kings (mGF% 52.6%) in 2012. Their first- and second-round upsets are still significant, but mostly because of the quality of the Canucks (53.7%) and Blues (53.9%).
- Penguins fans show no signs of letting go of either series, but their losses to the 2012 Flyers and 2013 Bruins wouldn’t have been shocking to the model. The 2011-12 Flyers were a strong team (mGF% 52.7%), and Pittsburgh’s win probability was just 53% in that series. In 2013, they faced a Bruins squad with a far higher mGF%, meaning they were actually underdogs in the series despite having home-ice advantage.
- The most competitive Stanley Cup Final of the past five seasons was actually last year’s Chicago-Boston series. The Hawks had a win probability of just 51.9% entering that match-up.
- Just missing the cut for the five most improbable series outcomes in the past five years: Ottawa’s (mGF% 49.8%) five-game victory over Montreal (53.4%) last season. Ottawa entered that series with a win probability of just 39.5%. Canadiens fans no doubt remember that their team badly outplayed the Senators for most of the five games, and then made awful personnel moves in the offseason to correct perceived weaknesses.
- Another honorable mention has to go to the 2008-09 Penguins. The Red Wings were moderately strong favorites to win that Final (probability 54.7%), but after winning the first two games, their win probability had jumped to 81.6%. As such, the comeback Pittsburgh made was very unlikely.
- How unlikely was Philadelphia’s comeback from a 3-0 series deficit against Boston in 2010? After Game 3, the Bruins had a 94.8% probability of winning that series.
Okay, without further ado, here are the five least likely playoff upsets from 2009 through 2013.
5. 2012 Eastern Conference Finals: New Jersey Devils (mGF% 48.1%) over New York Rangers (52.3%)
Everyone remembers the wild, upset-laden playoffs of the 2011-12 season, where an 8th-seeded team met a 6th seed in the Stanley Cup Finals; what’s often forgotten is that, despite coming within two games of winning the Cup, the Devils were not a particularly good team that year. In contrast to the New Jersey teams of the past couple seasons, the Devils were mediocre in puck possession, and with a 0.916 even-strength Sv%, they had some of the NHL’s worst goaltending. In the Conference Finals, they took on a Rangers squad with outstanding goaltending and the second-best record in the NHL, and at first, New York grabbed the upper hand. A 3-2 loss at home in Game 2 was bookended by 3-0 shutouts from Henrik Lundqvist, and after three games, the Rangers had a 75.6% probability of advancing to the Finals. Though Game 4 is probably best remembered for Mike Rupp punching Martin Brodeur and setting off a near-brawl and a shouting match between the coaches, it was an outstanding effort by the Devils, who outshot the Rangers in the course of building a 2-0 first-period lead, and cruised to a 4-1 victory. In Game 5, New Jersey was gifted a poor outing by Lundqvist, who surrendered three goals on six first-period shots, as well as the late game-winner after the Rangers fought back to tie; the Devils managed just 16 shots on goal in the game. Two nights later, the Devils blew a 2-0 lead and got outshot 21-12 over the final two periods, but Brodeur held strong, and an overtime winner from Adam Henrique completed the upset.
4. 2012 2nd round: New Jersey Devils (mGF% 48.1%) over Philadelphia Flyers (52.7%)
Did I mention that the 2011-12 Devils were not a very good team? Their victory over the Rangers was surprising, but arguably more impressive was their five-game win over Philadelphia the round before. The 2011-12 Flyers, as mentioned above, were a strong team, with a decent possession game and a scorching offense. Though they were above-average defensively, Ilya Bryzgalov and Sergei Bobrovsky had given them 0.919 goaltending that season, and that duo had been torched for 26 goals in the first round against Pittsburgh. Still, against a poor offensive team like New Jersey, it was reasonable to expect better results from the Philly tandem. The Devils came out strong in Game 1, taking a 1-0 lead behind a 15-6 shot advantage, but the Flyers would outshoot Jersey 30-11 the rest of the way and grab the victory in overtime. Game 2 played out in opposite fashion: Philadelphia took the early lead, but was smothered by New Jersey’s defending, putting just 11 shots on Brodeur in the final two periods, and four goals in the third gave the Devils the series split. When the series moved to Newark, the Devils took full advantage, taking Games 3 and 4 and outshooting the Flyers heavily. In Game 5, Brodeur was strong in net again, and the Devils seized a 2-1 lead in the first that they wouldn’t relinquish. For Philadelphia, the quick exit put a damper on the enthusiasm that had followed the victory over Pittsburgh, and led to much second-guessing of the team’s roster shakeups in the previous offseason. This is the only one of the top 5 unlikely upsets in which the expected winner was decisively outplayed.
3. 2009 1st round: Anaheim Ducks (mGF% 49.2%) over San Jose Sharks (53.9%)
Sharks fans have a lot of playoff disappointments to reflect on from recent seasons, but their six-game loss to Anaheim in 2009 is probably the worst, coming as an abrupt end to a President’s Trophy-winning season. For all the consistent excellence of the Thornton/Marleau era in San Jose, every team has had an Achilles’ heel that’s proven fatal in the postseason. In 2010-11, the Sharks were the best even-strength team in the NHL, but had an abysmal penalty kill; in their five-game loss to Vancouver that year, they were buried by allowing nine power-play goals. In 2008-09, their weakness was even-strength shooting: in spite of a dominating possession game and excellent goaltending, that season’s Sharks shot just 6.7% at 5-on-5. In their six games against Anaheim and then-rookie goaltender Jonas Hiller, San Jose scored just six even-strength goals. Hiller’s even-strength Sv% for the series was a stunning 0.966. The total shot differential in the series was a staggering 230-156. But all that territorial domination was pretty much moot after Anaheim took Games 1 and 2 in San Jose: at that point, the Sharks had just a 23.8% probability of winning the series.
2. 2010 1st round: Montreal Canadiens (mGF% 48.3%) over Washington Capitals (53.6%)
Without clicking on the link to this post, you probably could have guessed that this series would be mentioned. 2009-10 was perhaps Washington’s best chance to win the Stanley Cup with Alex Ovechkin; that year’s Capitals squad won the President’s Trophy behind the NHL’s greatest offensive juggernaut since the “Firewagon Hockey” days. The Canadiens, on the other hand, were a mess defensively, allowing more shots on goal than all but four teams that season; on offense, they had a miserable 7.6% even-strength Sh% to go along with taking fewer shots than all but five teams. Basically, they had a tremendous power play and, in Carey Price and Jaroslav Halak, a strong goaltending tandem, but little else. All expectations were that the Capitals would steamroll Montreal on their way to a second-round rematch with Pittsburgh; what actually transpired was expected by no one. Despite Washington putting 47 shots on Halak and taking a 2-1 lead in the third period of Game 1, a goal by (of all people) Scott Gomez sent the game to overtime, where Tomas Plekanec ripped home the winner 13 minutes into the OT. In Game 2, the Caps got off to a disastrous start, surrendering a 2-0 lead in the first eight minutes and going down 4-1 before roaring back to a 5-5 tie with 80 seconds left in regulation; a goal by Nicklas Backstrom seconds into overtime completed the comeback. In the next two games in Montreal, the Capitals tore the Canadiens apart. Though Halak is widely remembered as the series’ hero for Montreal, he was chased from the net in Game 3, and a four-goal third period in Game 4 gave Washington a commanding 3-1 series lead. At that point, their probability of advancing to the second round was 92.7%. Then, the Caps’ vaunted offense vanished inexplicably: though they put an astounding 134 shots on Halak in the final three games, they scored only three goals, and Montreal escaped with a stunning win. The Capitals, of course, were never the same: they would attempt to play a trapping system the following season under Bruce Boudreau, then fire him to try more of the same under Dale Hunter.
1. 2010 2nd round: Montreal Canadiens (mGF% 48.3%) over Pittsburgh Penguins (54.1%)
As upset as they get over losses like the 2012 Flyers series or last season’s Bruins sweep, it’s a little remarkable that Pittsburgh fans aren’t more bitter about this one. With a Fenwick Close of 53.7%, the Penguins were the second-best possession team in the NHL in 2009-10, and though my model wouldn’t have favored them against Chicago, the upsets of Washington and New Jersey would have cleared a fairly easy path back to the Finals. After dispatching Ottawa in the first round, the Penguins entered this series with a 65.8% probability of winning. Though many Penguins fans recall this as a series in which Montreal’s skaters “parked the bus” in front of Halak, Pittsburgh put 224 shots through to the Habs’ goal. Halak was once again superb, with a 0.956 Sv% at even strength. After a Game 1 onslaught, the Pittsburgh power play would only score four times in the remaining six games. Unlike the Capitals series, neither team really went on a roll in this series. Mike Cammalleri’s seven goals and Halak’s brilliance essentially allowed Montreal to hang on in the series until Game 7, in which a collapse by Marc-Andre Fleury (four goals allowed on 13 shots) sealed Pittsburgh’s fate. With that loss went perhaps Pittsburgh’s best chance at a second Stanley Cup in the Crosby/Malkin era.