How age, shooting percentage and ‘puck luck’ factor into future performance

Gerry Thomas

This work was completed as part of a larger project on age profiles and could not have been done without the valuable assistance, data and support of HockeyTech

At the beginning of the 2012-13 season, Kris Versteeg was a highly respected two-way forward whose resume included four 20-goal seasons and a Stanley Cup championship. There was little reason to think the 26-year-old didn’t have more great years ahead of him. As a result, the Florida Panthers rewarded him with a 4-year contract extension at an average annual value of $4.4 million.

The decision may have seemed reasonable to most pundits at the time, but then age and injuries took their toll and Versteeg bounced around among four different teams over the next four seasons.

Now 30, the former star came back to the NHL via a tryout contract in September after failing a physical with SC Bern in the Swiss League.

While injuries are difficult to predict, age comes for us all. Exactly when can vary among players, but there are some patterns to how players age, and it’s a mistake to believe there’s a “one size fits all” approach.

More skilled players, for example, tend to peak later and decline more slowly.

What complicates these patterns are the now widely recognized fluctuations in shooting percentages that can lead to players having abnormally good or bad seasons just due to “puck luck.” For example, many analysts will often take note if a player with a career average shooting percentage of 10% suddenly scores 30 goals one season based on 20% shooting. Chances are his shooting percentage will drop the following year and he won’t repeat those statistics.

This concept is what statisticians call “mean reversion”, which is a fancy way of saying that, in the long run, everything tends toward an average number even if there are short-term ups and downs.

Moreover, it’s not just a player’s own shooting percentage that can influence his point totals. Those of his linemates also have a big impact on the number of assists he gets. If you make a bunch of outlet passes that end up being second assists on your linemates’ lucky goals, you’re likely a good candidate for a disappointment the next season.

It’s all well and good to acknowledge these effects, but it’s another to quantify their impact accurately.

In order to do that, we looked at every single forward who played more than half a season since 2005-06 and tracked how his offensive production changed the following season.

We then considered how these changes were influenced by things such as age and mean reversion in individual and linemate shooting percentages (as well as a lot of other variables).

As it turns out, analysts who assume a player’s shooting percentage will always revert to his career average are missing a very important detail, namely that a player’s shooting percentage follows its own age curve.

Younger players tend to have lower shooting percentages that improve as they gain pro experience. Conversely, as players start to decline overall, one of the things that also goes is their shooting ability.

This means that a 23-year-old who shows a sudden jump in his shooting percentage may in fact be able to sustain that higher level in subsequent years. Equally, a 29-year-old who is performing at below his career average shouldn’t automatically be expected to rebound next season.

Now to the fun part.

We took every forward who played more than 41 games in 2015-16 and used our model to make predictions of what we could expect if he played a full 82-game season in 2016-17 (and yes we’re aware that guys will get injured, but good luck predicting that).

Below are predictions for four kinds of players:

(i) Stagnant Youth – Young players who won’t improve because their performance has been unsustainable

(ii) Breakout Players – young players who had bad luck previously and so will get the double kick of age and mean reversion

(iii) Veterans With Upside – older stars who will bounce back

(iv) Faded Stars – veterans who are in full decline mode

(Note that, because our model excludes guys with only one season under their belt, we’re not making predictions for last year’s rookie crop, which included some standout talents. So while we don’t doubt that Connor McDavid will contend for the scoring title while resurrecting the Oiler franchise, we’re not predicting an actual number for him).

Because age is so critical in this analysis, our model is kinder to younger players. So, for example, all of the players on our “Stagnant Youth” list are still going to have very good, and in some cases excellent, seasons.

For example, Artemi Panarin and Johnny Gaudreau will remain great; however, all signs point to last season’s performance having benefitted from some “puck luck”. So while they may improve simply due to additional experience, that effect will be counteracted by some mean reversion.


Name 2015-16 Age 2015-16 GP 2015-16 Pts 2016-17 Predicted points
A. Barkov 20 66 59 61
E. Kuznetsov 23 82 77 66
A. Panarin 23 80 77 70
J. Gaudreau 22 79 78 74
A. Wennberg 20 69 41 42
V. Trocheck 22 76 53 52
J.T. Miller 22 82 43 38
M. Scheifele 22 71 61 66
A. Duclair 20 81 44 40
N. Kucherov 22 77 66 66

The flipside of this is that younger players who underperformed last season – guys like Elias Lindholm, David Pastrnak, Nathan MacKinnon and Ryan Nugent-Hopkins – are poised for big jumps. They won’t necessarily surpass their “stagnant” peers, but just as most pundits are overestimating the previous group’s performance for this season, the “Breakout Youth” are heading in the opposite direction and so are generally being underrated.


Name 2015-16 Age 2015-16 GP 2015-16 Pts 2016-17 Predicted points
E. Lindholm 20 82 39 56
D. Pastrnak 19 51 26 51
N. MacKinnon 20 72 52 68
R. Nugent-Hopkins 22 55 34 59
S. Bennett 19 77 36 44
T. Toffoli 23 82 58 63
S. Monahan 20 81 63 69
T. Hertl 21 81 46 51
R. Fabbri 19 72 37 47
N. Ehlers 19 72 38 48

Meanwhile, older players will live out a simple fact that everyone should know: age is cruel, punishing, and rarely forgiving.

Finding vets with upside isn’t easy. And when you do, the reason usually is that a former elite talent like Eric Staal had a horrifically bad 39 point season that is expected to translate into an adequate 44 point campaign this year. Staal is never going to be the 30 goal 70 point talent he once was, but he’s not ready for his next career just yet.

In order to get a top 10 list for this group, we ended up mostly with guys who were expected to hold the line rather than improve. So “upside” for a vet means simply staving off his inevitable decline for one more season.


Name 2015-16 Age 2015-16 GP 2015-16 Pts 2016-17 Predicted points
E. Staal 30 83 39 44
D. Brown 30 82 28 32
M. Perreault 27 71 41 51
J. Pominville 32 75 36 42
D. Perron 27 71 36 43
N. Foligno 27 72 37 43
P. Hornqvist 28 82 51 52
P. Kessel 27 82 59 60
T.J. Oshie 28 80 51 53
P. Berglund 27 42 15 30

Last of all, our Faded Star group is made up of the types of players we might expect to see on a team’s unprotected list for June’s expansion draft. These are exactly the kinds of players their current teams would know are done and who an expansion team might optimistically imagine will have a renaissance in Las Vegas.


Name 2015-16 Age 2015-16 GP 2015-16 Pts 2016-17 Predicted points
H. Sedin 34 74 55 48
J. Ward 34 79 43 34
D. Sedin 34 82 61 51
C. Soderberg 29 82 51 42
L. Stempniak 32 82 51 42
B. Wheeler 29 82 78 70
J. Spezza 32 75 63 61
T. Vanek 31 74 41 37
S. Hartnell 33 79 49 43
Both Sedins, who have done everything as the identical twins they are so far,  appear to be joining each other for a permanent hiatus from elite scoring. The same goes for Scott Hartnell, who has more than 300 career goals and three consecutive 20-plus goal seasons. And while Thomas Vanek has gotten off to a strong start in Detroit early this season, we’d be very surprised if it lasts.

Predicting the future isn’t an easy business. But if you’re making player personnel decisions for any NHL team – or assembling an expansion team in Las Vegas – understanding the complex interaction between the dual phenomena of age and mean reversion is one way of making better predictions.

The Department of Hockey Analytics employs advanced statistical methods and innovative approaches to better understand the game of hockey. Its three founders are Ian Cooper, a lawyer, former player agent and Wharton Business School graduate; Dr. Phil Curry, a professor of economics at the University of Waterloo; and IJay Palansky, a partner at the law firm of Armstrong Teasdale, former high-stakes professional poker player, and Harvard Law School graduate. Please visit us online at

Jordan Hill is a masters student in the economics program at the University of Waterloo and a former Jr. A hockey player.

HockeyTech ( is a worldwide leader in providing hockey-related technologies, analytics and information services. HockeyTech was founded in 2013 by Stu Siegel (technology entrepreneur and former Florida Panthers Managing Partner/CEO) through a series of acquisitions. While HockeyTech is a new corporate identity, its brands have been providing cutting-edge solutions to the hockey world since 1998.  HockeyTech brands include RinkNet, ISS Hockey, HockeyTV Website, LeagueStat, and NEXT Testing.

This article originally appeared on