With the Christmas holiday passed and New Year’s Day approaching, it’s the time when hockey’s talking heads are looking at the standouts in this season’s rookie crop and making bold predictions based on years of hard earned “wisdom” that seem better suited to Halloween: BEWARE THE ROOKIE WALL.
Analysts and commentators love a good tale, often regardless of its accuracy. And so the Rookie Wall is a go-to as we hit the NHL’s mushy middle of the schedule.
In their 20 Fantasy Thoughts a couple of weeks ago, Sportsnet writers Ian Gooding and Steve Laidlaw wondered whether Arizona’s Max Domi, who is currently second in rookie scoring, will see his offensive production drop off. And back in March 2010, NHL.com Staff Writer John McGourty offered an explanation as to why some rookies hit the wall and others don’t: “The rookie wall differs for players who came to the NHL from college and those who took the major junior route. College teams play about 40 games. The Western Hockey League has scheduled 72 games the past few years while the Ontario and Quebec leagues play 68.”
So let’s get this out of the way right now: There is no rookie wall.
Now before we get out our slide rules and the old time commentators dust off their pitchforks, let us clarify. Many high-flying rookies do slow down in their offensive production as the NHL season progresses, and when people talk about “the wall” they tend to focus exclusively on those examples.
Consider the case of FilipForsberg.
When the Predators winger, then 19 years old, was decorating his Christmas tree last season, he seemed all but certain to win the Calder Trophy. On December 11, he was sitting pretty with 27 points in 27 games—on pace for an 82-point season. Meanwhile, Flames rookie Johnny Gaudreau had 21 points in 28 games—a respectable 62 point pace.
Moreover, Forsberg had played 74 total games during the preceding regular season, including 13 with Nashville and 14 for Team Sweden between the World Junior Championships and Olympics. Gaudreau played only 48, including one with Calgary and seven with Team USA at the World Championship.
wobbled to the finish line with a mere 63 points in 82.
Like Gaudreau, Ottawa’s Mark Stone apparently didn’t get the memo on what his season was supposed to look like either because he too improved, going from a mediocre 11 points in his first 25 games to a near point per game pace that tied him with Gaudreau’s 64 in 80.
Before we are accused of “cherry picking” a couple of outlier examples, we need to point out these “exceptions” aren’t as rare as you’d think.
We looked at the top 10 scoring rookie forwards as of December 11 in each season starting with 1995-96 (excluding the canceled 2004-05 campaign and the 2012-13 lockout-shortened schedule). Where there was a tie for the 10th spot, we included the extra guys.
That gave us 188 forwards over 18 seasons.
The first thing we had to figure out was how we actually determine what “the wall” looks like. You might be tempted to just look at total points for the rest of each season, but that doesn’t account for the effect of injuries.
TakeDerickBrassard of the Blue Jackets, for example, who was off to a very nice start by December 11, 2008, leading all rookie forwards with 22 points in 26 games. He finished the season with only three more in a mere five games after suffering a dislocated shoulder in a fight with James Neal that required season-ending surgery.
So while Brassard’s nice rookie campaign suddenly turned into an afterthought, it’s clear the only “wall” he hit was Neal.
To eliminate the impact of injuries, we decided to look at points per game (PPG) for the rest of each season while keeping in mind a bunch of external factors, all of which have nothing to do with a player’s individual abilities, that affect his PPG over the course of the season. Among them are changes in linemates, in-season switches of coaches and systems, use in different situations (e.g. offensive zone starts vs. defensive zone, penalty kill vs. power play, etc.) and perhaps most important, the fact that opponents get more time to study a rookie's strengths, weaknesses and tendencies, and plan for him accordingly.
These are all valid considerations. But the impact of them (and others) are already partially baked into the statistics for the 188 rookies we looked at.
The chart below shows how they did after December 11. Specifically, we calculated the difference between each player’s PPG up to December 11 and for the rest of the season. We put the players into “buckets” of 0.1 PPG each and considered the percentage of all 188 that fell into each bucket.
So, for example, a player who scored 0.05 PPG above his December 11 total for the remainder of the season would fall into the same bucket as one who scored 0.06 above his December 11 total.
As you can see, most of the top rookies did experience a drop-off in their production, with approximately 40% of them experiencing a decline of between 0 and 0.2 PPG and another 30% doing even worse.
As a group, the average player among our top 10 rookie scorers dropped 0.1 PPG from his December 11 total.
If there were a big, seemingly insurmountable “wall” standing in the way of rookies’ productivity as the season progresses you’d expect to see the majority of the top one experiencing big-time productivity drops. That just isn’t what happens.
To be sure, like Forsberg, some rookies’ scoring does drop significantly. But all too often commentators and analysts ignore the ones whose productivity significantly increases, like Alex Ovechkin and Sidney Crosby (the top two freshman scorers of the past 20 years), who increased their PPG from 1.14 and 1.03 up to December 11 to 1.40 and 1.38 respectively for the rest of the season.
Or Pavel Datsyuk, who came into the league in 2001 as a smallish 23-year-old center who was picked by Detroit in the sixth round three years earlier, and produced eight points in 30 games as of December 11. At that point Datsyuk was ranked 12th among rookie forwards, but behind such journeymen-in-waiting as Kris Beech, Toby Petersen, Krys Kolanos, Mark Bell and Kristian Huselius. In the remaining 40 games he played that season, Datsyuk suddenly found his scoring touch and potted 27 points.
Of course, nobody ever notices a rookie who goes from AHL worthy production early in the season, has an impressive run in the last half, and still ends up with only 35 points total.
But if there’s a rookie wall, what do we call Datsyuk’s case: a rookie surge? The fact is that commentators tend to see a wall where what they’re really seeing is mean reversion.
As any fan knows, points don’t come in a tidy and even distribution over the course of the season. Rather, all players, rookies and veterans alike, are prone to streaks and slumps. Over the course of an 82-game schedule, production tends to even out and move toward an average reflection of the player’s true talent.
The top 10 rookie scorers to whom the media is actually paying attention include many guys who generate the majority of their points early on, and they eventually fall off. Meanwhile, guys like Datsyuk, who didn’t get their points at the start of the season and are ignored as a result (and also aren’t in the top 10 as of December 11 so aren’t in our data), see their “puck luck” (which includes factors beyond simplly getting favorable bounces) improve later on.
However, regardless of what’s happening to rookies, we know that mean reversion isn’t unique to them.
Again Forsberg is a highly useful example here.
, with 62 points in 82 games.
Regardless of the specific causes, whatever factors affected rookie Forsberg’s production seem to have affected veteran Ribeiro’s in lock step.
So was Ribeiro’s production collateral damage from the rookie wall, or was Forsberg’s beholden to that of an aging linemate with talent and well publicized off-ice issues?
Or is it simply that neither was a point per game talent at this stage in his career but closer to the 50-60 point range (about where they’re both trending this season)?
Having disproved the existence of a rookie wall, we wanted to go the next step to see if we could tell which of the top 10 rookies were likely to regress or improve, and by how much.
We considered a bunch of different variables, including a player’s age at the start of the season, his number of games played during the previous campaign, shots, shooting percentage, power play points, shorthanded points, plus/minus (note to fellow analysts: please don’t hate us for even considering it), goals, assists, and of course points per game up to December 11.
As it turned out, among all the variables we considered, the best predictors of a top rookie’s production for the rest of the season are his shots per game, power play points per game, and points per game up to December 11.
If you’re looking for candidates who are most likely to sustain their production, find the guys who generated more shots and power play points so far.
Conversely, if you’re looking for guys who are going to hit this imaginary wall, don’t look for something that’s unique to them as rookies. Instead search for the kinds of things that folks in the analytics community would tell you suggest a player is due for some mean reversion, whether a rookie or a grizzled veteran.
Specifically, look for the guys who don’t generate a lot of shots, don’t get much power play time, and relied on a high shooting percentage to get their points in the first third of the season.
That’s bad news for Anthony Duclair, who had an absurdly high shooting percentage of 24.2% as of December 11 but generated relatively few shots (33, or 1.18 per game), and potentially good news for
, whose shooting percentage was a mere 9.8% while he was generating an average of 3.2 shots per game. (If you’re looking for a point of comparison,
, who’s never met a puck he didn’t want to fire at a net, had a walloping 5.0 shots per game at this stage of his rookie season, but the average top 10 rookie forward's average is only 1.9.)
So what about youth, inexperience and magic pixie dust?
As noted above, most commentators who talk about the rookie wall usually tell some story about younger players who are physically immature and can’t handle the physical grind of an NHL season, and some add that those who played in leagues with fewer games the season before will fade as they approach game number 82.
Neither of these fables turned out to be true.
It turns out that if two players have the same PPG on December 11, the younger one is slightly more likely to sustain it.
Moreover, the number of games played the previous season does a horrendous job of predicting production for the rest of the current one. The correlation is both weak and—oddly enough—negative, meaning of our 188 players, the ones who got into fewer games the previous season tended to fare better in the second two-thirds of the NHL season than those who played more games the previous year.
In this regard, the comparison of Forsberg (74 games, 13 of them in the NHL) with Gaudreau (48 games, one in the NHL) was not unusual.
Surely there are specific situations that may give some clues about which players are going to break down. For example, one who worked through a nagging injury over a long AHL campaign might end up playing through it in his NHL rookie season and eventually fade as a result of the pain. Meanwhile, a player who got into fewer games while playing NCAA hockey might arrive fresher at camp and better positioned to manage the NHL grind.
But if specific rookies crack under the grind, let’s not pretend that it’s something wildly different from what veterans go through. Most NHL players are going to feel worse in March than October, but not all will see declining production.
The table below shows the predictions our model, based on points per game, shots per game, and power play points per game, makes for this season’s top 10 rookies.
While Chicago’s Artemi Panarin should continue to lead all rookie scorers, his production will likely drop off—unless you assume that linemate Patrick Kane has another once in 20-plus game point streak in him. Meanwhile, when Connor McDavid returns from injury he is not likely, based on his first 13 games, to maintain his near point per game pace. That said, 13 games is a small sample, and McDavid very well may be special enough that the normal rules don’t apply to him.
If you’re looking for players whose performances are likely to make their coaches question whether they might need some time in the minors or the press box, guys like
seem to be closest to that territory.
Like most conventional wisdom in hockey, the rookie wall seems to be based on commentators’ vague impressions or excessive focus on a few specific cases rather than something as dull as actual facts. More important, to the extent there is a mild slowing among the top scorers, the ones who are generating fewer shots, haven’t generated points on the power play, and have unusually high shooting percentages are likely to suffer most.
If you’re in a fantasy pool and looking for the guys on whom you should “sell high” those should be your first candidates.
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 (@phil_doha), a lawyer, former player agent and Wharton Business School graduate; Dr. Phil Curry (@ian_doha), a professor of economics at the University of Waterloo; and IJay Palansky, a litigator at the law firm of Armstrong Teasdale, former high-stakes professional poker player, and Harvard Law School graduate. Visit us on line at www.depthockeyanalytics.com