The NHL starts similarly every season. Some goalie races out to a hot start and his team wins a bunch of games on his back. Some stumble out of the gate, and a lot of losing happens while fans and management wonder what they are doing to cause the problem.
You would think nobody would be able to forget the sparkling start Tuukka Rask had to last season. He posted an 11-2-0 record, earning the Boston Bruins 22 points through his first 13 games. He made 331 saves on the first 350 shots he faced to post a .947 save percentage. Unfortunately, I would imagine Boston fans don’t really want to remember the rest of his season where he struggled, posting a .906 save percentage the rest of the way going 26-18-5.
At the other end of the spectrum was Frederik Andersen, who posted an .898 save percentage through his first 12 games in a Toronto Maple Leafs uniform while going 6-3-3 to earn 15 points. He managed to right the ship through the remainder of his season, posting a .922 and going 27-13-11.
It is well known by those in the NHL analytics community that goals scored and allowed are a relatively poor predictor of future goals for and goals against. This is one of the main reasons analysts rely on other metrics such as shot attempts or expected goals. Goal scoring and prevention are volatile metrics in short time frames. We know that players get hot or cold over small samples and long-term results often don’t align with those shorter stretches of play.
This season a trio of Canadian franchises have been reaching towards the panic button as they watch their starting netminders struggle through the early going. Andersen has posted a remarkably similar opening 12 games to last year, with a .901 save percentage and 7-5-0 record. Cam Talbot in Edmonton and Carey Price in Montreal – who finished fourth and third in Vezina voting last year, respectively – have posted similarly atrocious starts. Talbot has an .904 save percentage through his first 10 games with a 3-6-1 record, while Price has an abysmal .877 save percentage through his first 11 games and a brutal 3-7-1 record.
This is where we can delve a bit deeper. While all three goaltenders are performing well below expectations, it is worth asking if part of that is based upon the expected save percentage (xSV%) on the shots the teams in front of them are giving up. If we look at how their team’s defensive results compare to the NHL average, we can see that in this instance, team defence doesn’t necessarily explain why their numbers are abysmal.
If we look underneath the hood a tad we can see that each of the aforementioned goalies is struggling with shots of a specific type, which can inform what we expect to see happen in the future.
Low-danger save percentage (LDSV%) is the most volatile of the three danger-zone save percentage values. This means the results shift most randomly and uncontrollably by a specific netminder. These are long-distance shots that tend to carom in off deflections, go through screens, or result from odd puck movement as the shot travels from a long way out.
Virtually all NHL goaltenders will see their results settle within a very narrow range at the end of the season. Every goalie to play over 2,000 minutes since the 2010-11 season has posted an LDSV% above .951, with 75 per cent of them exceeding .974 and the top 25 per cent being at .982 or above.
Talbot’s current LDSV% of .991 would rank in the top eight out of 214 goaltending seasons since 2010-11. While it is entirely possible that number remains high, as we can see from his past three seasons, his LDSV% is typically closer to the league average of .979.
Andersen’s low-danger results are similarly abnormal, both in comparison to the rest of the league, and his own performance the past three seasons. A .949 LDSV% season would be the lowest season on record for a regular starter, so the probability of this being his result at the end of the year is amazingly low. We should presume this number rises to at least the .974 range attained by 75 per cent of all starters. Beyond that, there isn’t a lot to worry about with Andersen as his mid-danger and high-danger results so far are both reasonably close to his outcomes the past three years in both Anaheim and Toronto.
Talbot and Price are both having problems with respect to mid-danger shots, posting results well below their personal norms. Talbot usually thrives in this category, regularly posting amongst the top 10 numbers in the NHL. His .879 MDSV% is bound to improve as the worst result for any regular NHL starter in the past seven seasons is .884. Seventy-five per cent of NHL starters can hit .915 in MDSV%, so we can presume he at least gets towards that level.
Price has further to go for his mid-danger save percentage to improve, and it will, but I’d generally work under the assumption that he brings things up to the .884 range before targeting the .915 I mentioned for Talbot.
The main area of concern for Price and the Habs is definitely his high-danger save percentage. High-danger goals make up the majority of goals scored in the NHL, despite accounting for less than 25 per cent of the shots a goaltender faces. The spread in HDSV% is the largest of the three tiers, and thus HDSV% is the key distinguishing factor in what makes or breaks a legit NHL starter. Almost every goalie in the league can stop shots from the point with a fairly high degree of repeatability. What is hard to do is stop NHL snipers who are firing away from point blank range. You will note that of the nine prior individual seasons shown for the three goalies we are discussing, only Andersen’s 2015-16 had an HDSV% that fell below NHL average.
Carey Price’s historical results in particular are sparkling in this regard, which is what makes it so worrisome that he is struggling mightily this season. The scary part for Canadiens fans is, while a .750 HDSV% is abnormally low for Price, it remains above some of the worst seasons posted by an NHL goalie with 2,000-plus minutes since 2010-11. There have actually been nine seasons posted in the past seven years that are equivalent or worse – over a full year. And while one might expect a list full of goalies nobody has ever heard of, a name or two might surprise.
Pekka Rinne posted the lowest HDSV% in that time frame in 2015-16 (.721), and the third-lowest in 2016-17 (.738) – right before he led his Nashville Predators on a crazy run to the Stanley Cup Final. Jimmy Howard posted the second-worst HDSV% result in 2014-15 (.731), which happened to be Mike Babcock’s last year in Detroit. Interestingly, in all three of these examples, the team the goalie was starting for still managed to make it into the playoffs.
I would anticipate Price’s results on high-danger shots to improve, but how much it does so is open to debate. Raising his full-season result above .770 is a realistic target to aim for, and the league average of .791 isn’t out of the question. I remain doubtful that he can push himself to the .820s he has posted historically.
So if we project using the results discussed, where do these three ‘keepers end up to finish the season? I anticipate Price will end up with a full-season save percentage slightly above .900 at the minimum. It may go higher, but his usual .915-plus territory is likely already out of reach. Andersen’s early-season dip has likely brought his seasonal results to below what Leafs fans would expect, but he’ll likely post a mediocre .912 save percentage or better when all is said and done.
Of the three, Talbot’s results are likely to look the best at the end of the year, largely because his HDSV% remains in the upper-end territory we are used to him providing. He should be able to post a .917-plus save percentage and if he runs hot later on this year, it could go higher.
Goaltending is fickle, but teams and their fans can be even more so. It is best to ride out the storm and let the guys they are paying massive amounts of money to sort themselves out rather than panicking and doing something they regret. Long term, all three of these goalies are likely to revert towards their historical performance.