Which Players are Primed for a Downward Percentage Regression in 2020-21?

Whose stats got inflated by outlier shooting percentage?

One of the basic benefits of analyzing hockey statistically - even if you choose to disregard all player evaluation models like WAR and RAPM - is it allows us to recognize and anticipate the areas where the sheer luck inherent to the sport might be creating outlier outcomes. I explained in an earlier piece just how pronounced the luck factor is when it comes to things like individual and on-ice shooting and save percentage. The impact of true, repeatable skill on these numbers is very slim, and goals, assists, plus-minus, and even the eye test can be misleadingly inflated, giving us unrealistic expectations for future performance.

In this article I’m going run through some names who you should expect at least some degree of regression from next season. Just because a player had some positive shooting luck doesn’t make them a total fraud or bad or anything like that. But fans can be extremely punishing to guys who don’t live up to lofty expectations, and I want to do my best to make sure that people aren’t setting themselves up for disappointment. I’m not a fantasy hockey player myself, but this information might also be valuable to you if you’re looking to make sure you don’t pick up a player who might be drafted too early based on an unexpected bump in point totals last season. I will follow this up with a list of players who are primed for a bounceback this weekend.

Shooting % stats come from NaturalStatTrick, expected goals stats come from EvolvingHockey.

Individual Shooting

Let’s start with individual shooting percentage. 28 players who played at least 500 minutes last year shot over 17%. Read the names on the list and estimate how many of these guys you figure will hit 17% or higher again next season.

Based on historical data, the answer is probably around three or four. Last year, it was four out of thirty-one.

Some guys who almost certainly won’t are pretty obvious - it goes without saying that you probably shouldn’t bet on Artem Anisimov, Colton Sissons, or Scott Laughton in fantasy. But there are plenty of players worth keeping an eye on. Here are some especially noteworthy guys whose 2019-20 scoring results were likely over-inflated:

Sebastian Aho

Aho scored a career-high 38 goals, tied for 7th in the league. This was very good news for the finishing-starved Hurricanes, but they probably shouldn’t bank on a repeat. He was on pace for over seven goals above expected and shot four points above his previous career high.

Mika Zibanejad

Zibanejad’s insane per-game rates last season have some arguing that he should be considered a top 10 or even top 5 centre in the NHL. I am sad to report, however, that Zibanejad is not likely to replicate his 58-goal pace this season - players don’t tend to shoot 19.7% in all situations again (if ever). At his previous shooting percentage, he would’ve been on pace for around 38 goals this year, and that seems like a much more reasonable (and still impressive) expectation.

Bryan Rust

Rust was on pace to score 40 goals and 83 points in a full season, which seems insane considering he’s Bryan Rust. While an established top six and PP role should get him solid numbers, last season looks like a pretty huge aberration.

Jean-Gabriel Pageau

Both the Ottawa Senators and Pageau had an objective last season: inflate Jean-Gabriel Pageau’s value as much as possible. And they both succeeded above their wildest dreams. For the Senators, it meant playing him 19 minutes a night as their #1 centre so they could get a haul of draft picks at the deadline, which they did from the Islanders (a 1st, 2nd, and 3rd). For Pageau it meant hitting the jackpot in the shooting lottery to the tune of a 31 goal pace so he could get a multiyear contract, which he did from the Islanders (6 years at $5 million). Everybody won here - even the Islanders for the time being at least, since he shot 17.8% in the playoffs as well. I admit I am deeply skeptical about how sustainable those results are based on his track record:

On-Ice Shooting

Even more random than a player’s individual shooting percentage is on-ice shooting - while there are ways a player can elevate his linemates’ shooting through passing or screening, they are much murkier than the direct impact he has on his own finishing. On-ice shooting boosts a lot of stats, like assists, points, +/-, and even EvolvingHockey’s WAR (which uses RelTMGF as the main input for offensive GAR), often leading to unsustainable results and unrealistic expectations.

Here’s the list of players with a 5v5 on-ice shooting % above 11% in 2019-20. Some real stars here - how many do you figure will be on this list next year?

The answer is probably one or two. Last year only Kuznetsov was a holdover. So let’s take a quick look at some guys who had some good fortune this year (other than the ones we already looked at).

Ryan Ellis

Ellis finished third in the league in EvolvingHockey’s WAR model with 4.1 in just 49 games - on pace for 6.8, which would have been the 2nd-highest ever recorded and the highest for a defenceman by 1.9. Suffice to say, this was probably just a bit inflated to say the least. Ellis is a superb defenceman but it’s probably fair to say that he was not responsible for his team being on pace to score almost 33 goals above expected at 5v5 with him on the ice. Intriguingly, something similar happened in his injury-shortened 2018 campaign as well - apparently my guy loves starting off hot.

Artemi Panarin

Panarin is a superstar who had a very good case to be the MVP this season due to the immediate even strength impact he had on the New York Rangers. Expecting a repeat of his incredible season could set Rangers fans up for disappointment, though. While Panarin’s top-three playmaking ability undoubtedly elevates the finishing numbers of his linemates, he is almost certainly going to see at least some regression this time around. He’s more likely an 80-90 point player than a 115 point one.

Jonathan Huberdeau

Huberdeau has scored at a 92 point pace in each of the past two seasons, which seems like the most sustainable bet imaginable. But the way those points came this year was very odd. As I noted in my first-half vs. second-half analysis, Huberdeau’s elite underlying numbers in 2019 fell off a cliff in 2020. His on-ice shooting did too, but it didn’t fully regress as he ended the season at 11.4%. I think it’s fair to expect that we’ll see something more in line with his previous two seasons at 5v5, and when it comes to total points it might depend what kind of PP the Panthers have next year.

Leon Draisaitl

I hestitated about including Draisaitl here but I will just as a point of conversation. Drai is an elite finisher and playmaker, and it should be expected that he will see lots of shots go in when he’s on the ice. That being said. I think it’s fair to say that at least some of that on-ice shooting bump comes from the benefits of playing with Connor McDavid, with whom he can generate high-percentage plays like odd-man rushes and cross-seam one-timers. 29 goals on 21 expected goals is kind of what you’d expect from that line. But this season, the line of Ryan Nugent-Hopkins, Draisaitl, and Kailer Yamamoto scored 28 goals on 16 expected goals. That’s a great line, but it seems like a bit much to expect that kind of absurd finishing from that trio again.

I say this because the Oilers - wisely, in my opinion - seem to be favouring splitting the two up. Overall, I expect this to pay dividends for Drai’s overall impact, especially on defence. But I think 110 points is a lot more reasonable as a baseline expectation for the type of production you’ll get from Draisaitl than the absurd 127 he was on pace for last year.

Conclusion

Many of the players listed here are among the league’s elite, and should be expected to perform very very well this season. Pointing out these percentage anomalies isn’t about throwing scorn upon a player’s 2019-20 performance or saying that their results were pure luck, it’s just about setting expectations reasonably so we don’t blame them if their numbers fall back to being just really really good instead of absurd or transcendent. In a 56-game season, percentage-based randomness will be even more pronounced than in a regular 82-game schedule, and maybe some of these guys will be fortunate enough to keep the party going in a smaller sample. But just don’t complain that they’re “disappointments” if they don’t!