March 16 2015 12:10PM
Photo Credit: Robert Mayer/USA TODAY Sports
Michael Hutchinson has cooled off from his hot start.
Since the All-Star break Hutchinson has posted an ugly save percentage of .879, the lowest of any other NHL netminder aside from Anton Khudobin.
So how much should the Jets worry about their young goaltender?
Jets fans were genuinely excited for the emergence of Hutchinson, and with good reason.
Hutchinson was coming off of a 0.923 save percentage in the AHL, bringing his career numbers to 0.917. When the pressure rose, he stopped 93.8 percent of shots faced and carried the St. John's IceCaps to the Calder Cup championships. He also posted a phenomenal 0.943 in a three game cup of coffee with the Jets.
Hutchinson vindicated those hopes by being near unbeatable in the early going this year. The 25-year-old netminder started off the season with a 0.935 save percentage in over 20 games prior to the All-Star break, which was the best in the in the league among goalies starting at least 20 times.
And then Hutchinson's form collapsed.
In the long run though, the Jets should not be worried.
Michael Hutchinson has been a strong performer at every level in almost every season of his career.
Despite the cold streak, Hutchinson sits at 0.916 NHL career save percentage, just over the average typical of a NHL starter. While Hutchinson's good games were concentrated early in the season and his poor games at the back, Hutchinson still has prevented about the average amount of goals one would expect if he were a bonafide starter.
Most importantly, his performance remains superior to that of any starting goaltender the Jets have employed since the move.
Looking at rolling sets of five consecutive games, Hutchinson still has the better 5-on-5 save percentages more often than Pavelec:
So while he's been awful of late, Hutchinson has still been a legitimate starter for Winnipeg this season.
When you have a rookie netminder who posted above average numbers in both junior and the AHL hold a 0.916 save percentage after 39 games and over 1000 shots in the NHL, you can hold some optimism for the long-run future.
What about the short-run?
Here's the rub. The short run matters for the Jets... a lot.
The Jets are in a middle of a tooth and nail battle for the playoffs. They sit just inside with thirteen games to go and the margin for error is thin.
Since the All Star Game, Hutchinson has been below replacement level. Ondrej Pavelec has not been much better, pacing close to his career norm with a 0.907.
The Jets will likely need something around a 0.913 save percentage or better to make the big dance if they continue their strong performance in the shot metrics. Neither goaltender has been giving them that lately.
The good and bad news is a lot of small sample save percentage is largely "luck" driven.
The relationship between a goaltenders true-talent level and their performance over a random fifteen game sample is loose at best. A 0.920 true-talent All-Star is expected to be out performed by a replacement level 0.900 true-talent one out of six times.
There are underlying drivers to hot and cold streaks. We see streaks occasionally persist longer than what would be expected from "true" randomness within a probability distribution due to the immeasurable human factors like confidence. The human factor is real; these athletes are real people. We're not meaning to ignore that when we say that, even so, these streaks do not last forever and in the big picture model pretty close to random selection.
What does this mean?
Look back up to the density traces for Hutchinson and Pavelec. The next five games for either goaltender will likely fit anywhere in that distribution, with the higher the curve, the more likely that result. If you include the human factor, confidence of both struggling lately may drag the probability down to the left a bit more.
All streaks snap eventually. If poor games were the only driver of future performance, streaks would last forever.
Cam Charron showed that for James Reimer, the netminder actually performed better after poor games than strong games. He then shows that the difference is about the same when he arbitrarily compares even versus odd numbered games. In the article Charron mentions that Reimer's results are not an exception to the rule when compared to "20 or so" other goalies.
Playing the "hot-hand" over the long run seems inferior to using the superior overall netminder (depending on rest days).
In a macro sense, while there are underlying reasons for streaks, the overall results closely mirror randomness.
Humans by nature look for patterns in randomness. It is why the shuffle function on the iPod was designed to be not be completely random to seem more random.
If you were to fill two cards, one with falsified coin tosses and the other with a real one, an observer asked to select which one was the real experiment will usually pick the made up one. If asked why, they tend to feel the real one had too many large clusters of consecutive heads or tails to be random.
We're about 2000 shots shy from getting a real confident look of Hutchinson's true-talent level. Still, Hutchinson has performed well on average. He has also out performed Pavelec on the whole, despite not playing well as of late. He also has strong indications of high talent due to performance in lower leagues.
While we are not as confident about Hutchinson being the superior netminder choice as we would be if we were given a 3000 shot sample on Hutchinson, all current indications are that he still gives the Jets the best chance at making the playoffs.
The caveat though is the human factor. Often the human factor is falsely applied to make sense of randomness, but it is a real thing. Hutchinson offers the Jets best chance, but they gain an even better chance if they can bring his confidence back up.
The Jets need Hutchinson, but they need a confident Hutchinson.
No matter which way you cut it, the Jets will need a smile and a bounce or two from the hockey gods if they are going to make the playoffs.