The primary reason that hockey teams make trades is that they identify a change they’d like to make and they see some kind of swap of assets as the best way to accomplish it. It could be that they want to add some depth for the playoffs, or lower the amount of money they’re spending, or any other kind of shift they think would benefit their roster. But there’s another reason that teams ought to explore the trade market that isn’t tied to any specific change they want to make, and that’s to take advantage of differences in what a player’s perceived value is compared to their actual value. Teams should be trying to find players whose current team under-values them (*coughondrejkasecough*) and acquire them at below market value. They should also be trying to figure out who on their own roster might be over-valued by other teams, and trading those players while their value is at its peak.
Perhaps this is too vague, so allow me to provide a more clear example. Let’s say that a team has a player who we’ll call Andreas Johnsson. And let’s also say that, for some reason, other teams might have a belief about how good Andreas Johnsson is, but that he is in fact not that valuable. In that situation, it’s quite likely that you could find a team that is willing to give you more for Andreas Johnsson in a trade than he is worth to you on the roster. You could, in all likelihood, then turn around and turn those assets into a player or players who help you even more than Andreas Johnsson would have. Think of it as an arbitrage opportunity.
In this article I’m going to explain precisely why I think that’s the situation the Toronto Maple Leafs are in right now. But first, a brief diversion.
A BIT OF HISTORY
Connor Brown played his rookie season in the NHL in 2015-16. He scored 20 goals and won over many fans with his energetic play and unexpected scoring punch. Back at the time, I suggested that the Leafs should consider trading him (I didn’t write an article like this so you’re going to have to take my word for it) because his value was at its peak. Many Leafs fans objected to the idea on the grounds that the reasons other teams would want him (his contract was inexpensive and he had just scored 20 goals) were the same reasons the Leafs should keep him. Over the subsequent two seasons his goal total fell to 14, and then to 8, and it seems like a foregone conclusion that he will be traded this summer to help ease Toronto’s salary cap crunch.
I don’t think it will be at all difficult to trade Brown, whose contract is still pretty team-friendly if he’s going to play in your top 9, but his value has surely fallen pretty substantially from where it would have been after his 20 goal rookie campaign. The Leafs could be in a better position today if they’d exercised a bit more foresight a couple of years ago.
Leo Komarov provides another great example. He also had a big year in 2015-16, scoring 19 goals in only 67 games, and representing Toronto at the All Star Game. A number of Leafs fans, including me, argued that the Leafs should trade him to take advantage of a likely surge in his value, which had been propped up by a shooting percentage that was nearly double his previous career total. Many fans again opposed the idea on the grounds that, having just hit a major new career high, Komarov was as valuable to the Leafs as ever. We know how that went; Komarov’s goal and point totals continued to decline throughout his last two years in Toronto, and he wound up finishing his career in Toronto as a healthy scratch in the playoffs in favour of . . . Andreas Johnsson.
Let’s snap back to the present. Even if it is the case that there are players who are sometimes overvalued, do we have any particular reason to believe that Andreas Johnsson is one of those players right now? In fact, we do.
Jonas Siegel had a good article at The Athletic recently exploring Johnsson’s season. Among the things Jonas noted was that Andreas Johnsson scored a very high rate of points at 5v5. From November 24 until the end of the season only three players in the NHL had a higher rate of points per 60 minutes at 5v5: Patrick Kane, Nikita Kucherov, and Johnny Gaudreau. Pretty elite company! But there are reasons we should be skeptical of Johnsson’s results.
Last year Andreas had a very successful rookie season by most measures. He finished with 20 goals in 72 games while playing under 14 minutes a night and getting limited powerplay time for most of the year. 16 of Johnsson’s goals came at even strength, which tied him with Mitch Marner for 3rd on the team in 5v5 goal scoring. Pretty impressive considering that Marner played nearly 50% more minutes. So far this sounds like an explanation for why Toronto should keep Johnsson, not why they should trade him. So what’s the catch?
The catch is this: Andreas Johnsson had a S% of 15.7% at 5v5 last year. That wasn’t actually the highest on the team (John Tavares and Tyler Ennis were both higher), but it is unusually high. The league average shooting percentage for forwards at 5v5 last year was 9.9%, so Johnsson was well above that number, and that’s something we should be wary of. While there are some players who consistently shoot substantially above league average (Steven Stamkos and Auston Matthews, for example), as a general rule players who have unusually high shooting percentages in one year will tend to regress towards the mean in subsequent years.
But don’t take my word for it: I’ve got math!
To show how unlikely it is for Johnsson to shoot so well again next year I’ve collected every forward who shot at least 14% at 5v5 over three seasons (2015-16, 2016-17, and 2017-18) and compared their results to the following season to see how well their shooting held up. In addition to 14% or higher shooting, I’ve limited my results to players with at least 75 shots so we’re not throwing off our conclusions by looking at players who had huge shooting percentages on a small number of chances. There were 20 such forwards in 2015-16, 24 in 2016-17, and 22 in 2017-18, which is a pretty consistent sample size.
First let’s take a look at how each group did as a whole in the season they shot at least 14% and the subsequent season:
|Year||Group S%||Next Year Group S%||Next Year League Avg S%|
(The raw data for my calculations was pulled from the excellent Natural Stat Trick. League average is for forwards only.)
You can see that these players collectively had a fairly substantial drop in shooting percentage from one year to the next. On average across all three seasons the forwards who shot 14% or higher in one season saw their S% fall by 4.6% the next season (the median drop was 4.8%). In fairness to these players I must point out that they did on average continue to shoot at a better than league average rate, so we do have some evidence that the group as a whole is comprised of above-average shooters. But broadly speaking they did see substantial decreases in their rate of converting shots.
Very few of the 66 players in my sample were spared. 7 of them saw their S% increase the following year (Mark Stone is included twice!), but the other 59 saw a reduction. 46 players declined by more than 3%. So while it is impossible to say with certainty that Andreas Johnsson’s 5v5 S% will decline substantially next year, it is by far the most likely outcome.
One objection you might have is that not all shots are equal. Sure, many players get lucky, but some players really are good shooters, and we should be taking shot quality into account. Thankfully there are a number of expected goal models that attempt to account for shot quality. I’ve decided to use MoneyPuck’s xG model for this article. MoneyPuck tells us that while Andreas Johnsson actually scored 16 goals at 5v5 last year, based on his shot quality we would only expect him to have scored 10.5, for a difference of 5.5 goals above expected.
That ranks Johnsson pretty high up in the league as far as goals above expected, but I think there’s a slight flaw there for our purposes, which is that the number doesn’t give us a good sense of the scale to which Johnsson outscored his xG. Think of it this way: If a player has an xG of 30 and scores 35, that’s a much smaller diversion than a player who has an xG of 5 and scores 10. The first player only score 17% more goals than expected, while the latter scored 100% more.
To that end, I download MoneyPuck’s player data from last year and calculated which players had the highest percentage of goals scored above expected at 5v5. I limited the results to players who scored at least 5 goals to prevent large outlier numbers from muddying the results.
By this measure, Andreas Johnsson outscored his expected goals by one of the widest margins in the league: 53%. That ranks Johnsson 30th out of the 325 players with at least five goals at 5v5 last year. He’s not at the very top of the league (that honour goes to David Perron), but it’s further evidence that Johnsson was scoring goals at an unsustainable rate last year.
STATISTICS THAT AREN’T GOALS
Having established that Andreas Johnsson’s goal scoring prowess is likely overstated by his results last year, I’m not sure I’ve yet established that the Leafs should explore his trade market value. After all, he could very well see a decline in his goal total but provide enough value elsewhere to make it difficult for the Leafs to part with him. So let’s take a look at some of Johnsson’s other results. As a point of comparison I’m going to include Kasperi Kapanen’s numbers in all the same categories.
There’s a fantastic tool called the A3Z (All Three Zones) Player Comparison Tool that shows where players around the league fall in comparison to each other in various neutral zone stats like zone entries and exits, as well as a couple of shooting and passing statistics. It’s measured in percentiles, so a player with a 95 is in the top 5% of the league, while a player with a 5 is in the bottom 5%. Here’s how the two players stack up:
You can easily see how Johnsson struggles in the neutral zone, and we know from past research that good results in the neutral zone are a major factor in terms of driving shot differential (ie. Corsi). Johnsson is OK but not great at entering the offensive zone with possession of the puck and quite poor at getting the puck out of his own zone. Kapanen, used as a basis of comparison, is one of the league’s better players in terms of getting the puck into the offensive zone, and is also better at creating offence overall.
Johnsson also fails to generate a particularly high rate of shots or shot assists. (A shot assist is a pass to a player who then shoots the puck.) One note of caution is that there are only 20 games worth of data for Johnsson in these statistics, which are manually tracked by Corey Sznajder. However, in aggregate the data suggests that Johnsson’s impacts are not particularly large in some important areas of the game other than shooting.
Andreas Johnsson’s goal total last year was likely inflated by shooting luck that we would expect to regress, possibly quite significantly, in subsequent seasons. Because of this, there’s a reasonably good chance that there are teams that see his high rate of goals and points, and would see him as having pretty substantial value. His status as a “rookie” last season might seem to suggest that he has a lot of room to continue to grow as a player. However, he will be turning 25 years old early in the 2019-20 NHL season, and we know that on average scoring peaks at age 24. These are reasons why another team might value him, but the Leafs should be wary on both fronts.
To be clear, I am not saying there is any particularly urgent need for the Leafs to trade Johnsson. So long as they can keep his cap hit down he’s still a useful member of the team. But since there seems to be a reasonable chance that another team will overvalue him, and since then Leafs could use the assets and cap space elsewhere (such as upgrading their top 4 defence), it would be wise to see if Toronto could find a team that places a high value on Johnsson now, when his trade value could be as high as it’s ever going to be.