Prior to his suspension Winnipeg Jets forward Evander Kane had only accumulated four goals and six assists in twenty games. There was a lot of talk about "poor play" and the two game prohibition only added fuel to the fire.
There are some out there screaming for Kane to be traded before his value tanks to nothing.
Is Kane's drop in scoring an indication of a downward trend or is this a cold streak where we should expect Kane to bounce back?
What do the numbers say about Kane's impact on the team?
Corsi (shot attempt differentials) and other shot metrics are not everything, but they are a large piece of the puzzle. They must not be ignored if the end goal is an accurate analysis of a player's impact. The biggest issue with Corsi is that a player's usage also impacts their Corsi percentage. One way to account for this is by looking at the player's trend in both Corsi and Corsi's major outside factors.
The usage factors are ordered in terms of their impact on a player according to most research.
First off we see that Kane's zone deployment has been increasing in difficulty. Each season Kane has started more shifts in the defensive zone relative to how many he starts in the offensive zone. For perspective, of the Jets top 9 players, Kane has gone from the primary offensive zone option to the primary defensive zone option. This should have a major negative impact in Kane's Corsi, but we see it has been trending upwards with the exception of 2011-12 to 2012-13.
Then there are the two smaller, but still legitimate, contextual nuances. TmCorsi% is Kane's linemates' Corsi% for their minutes away from Kane; OppCorsi% is the same but for Kane's opponents. The take home here is that in every season Kane's teammates do better -and opponents do worse- with Kane than without.
That is only Corsi though; I care about goals...
The point of Corsi is to determine the sustainability of low sample goal differentials, or predict better what future goal differentials will be like. Some may claim Kane as the exception of the rule, but there is little evidence that major exceptions exist. Also, the burden of proof would then falls to them anyways.
Regardless, with Evander Kane on the ice, the Jets have controlled 55 percent of even strength goals (typical NHL spread is 40-60).
The Jets are outscoring their opponents pretty severely with Evander Kane on the ice and his Corsi adds assurance that it isn't due to the Jets getting more than their fair share of bounces. That in itself is evidence that Kane is helping the Jets win and will continue to do so.
Ya, but he isn't the one getting the points...
Currently that is true.
The evidence above shows that Kane has indirectly impacted the Jets scoring in a positive manner; however, he has yet to have a large impact directly in scoring. While indirect factors are often grotesquely undervalued and misunderstood by many, Kane historically is a player who impacts scoring directly and is getting paid because of this.
Evander Kane is currently pacing with the Jets second best scoring rates in both power play and penalty kill, so the issue lies with even strength scoring. How different is his 5v5 scoring then?
Those are some huge drops in scoring. For 5v5, Kane currently has two goals and one assist. Historically he would normally have 5 goals and 5 assists at this point.
Kane's scoring pace has dropped by 60 and 80 percent in goals and assists. However, his non-blocked shot attempt production has only dropped by about 15 percent.
A decrease in shot volume could arise from either (or both) usage or the player. This decrease indicates Kane is creating less offense, but not nearly to the degree his scoring numbers have dropped.
Where has Kane's drop in point scoring come from then?
Goal scoring comes from a combination of shot volume and the percentage of shots that go into the net (duh).
For a player, goal scoring is a combination of their own personal shot volume and personal shooting percentage. Assists are a combination of the team's shot volume, the team's shooting percentage, and the percentage of the team's goals a player assists on.
History has shown that shot volume is mostly stable even at smaller samples, while percentages can vary greatly. Over time, these percentages tend to regress towards a player's historical mean (assuming that the historical sample is of significant size). This is true even for the best players, as shown here with Sidney Crosby.
To estimate the impact of one statistic changing, all one would have to do is change one statistic while hold the others constant. For goals scoring, this means holding one of shot volume or shooting percentage to career norms and then looking at the change with the other.
|TOI/GP||Volume (per 60)||Percentage||ΔGoals|
Statistics used in measuring are: iFenwick/60 as shot volume and iFenwickSh% as shooting percentage.
Paul Maurice reducing Kane's ice time by about half a minute a game causes nearly a negligible impact in Kane's goal totals. Combining the ice time factor and Kane's reduced shot volume only equates to about one goal less than his historical norm.
The bulk of loss goals has arisen from a severely reduced shooting percentage. Normally Kane scores on 6.1 percent of his non-blocked shot attempts, except this season he's only scored on 2.8 percent.
For assists, the process remains the same but instead using team goals rather than player goals, and adding that only assists on a percentage of the goals scored while on the ice.
The pattern is similar here. Again ice time and shot volume changes are only very minor factors in Kane's point production decrease. A change in shooting percentage of Kane's line mates has added to his low assist totals, but not by much.
Here the large difference is in IAP, the percentage of goals while a player is on the ice that they gain an assist on. IAP is not discussed that often, but behaves similarly to shooting percentages. It has large amount of variance and tends to regress heavily over time. It's not to say that a player cannot sustain above or below league average IAP but that a player with dramatically different IAP indicates that bounces haven't gone that players way.
Normally Kane assists on 35 percent of goals that have occurred while he is on the ice, but this season Kane has only assisted on nine percent.
Those percentages just show how he takes terrible shots / blind passes
This is not true. Those numbers do not tell you that. This does not mean that Kane does not take a lot of low quality shots or that Kane could not improve his playmaking ability. What it means is that these numbers are not low due to Kane's talent in those areas.
For one, even a blind squirrel finds a nut every once in a while. Kane's shooting percentage on non-blocked shot attempts and his IAP is low. Really, really low. It is below what you'd expect even if he was absolutely terrible in those areas Kane is often criticized for.
Secondly, we already know Kane was far better in these areas for 3000 minutes; 3000 minutes where Kane was already receiving criticism in regards to his shot location and playmaking. This past 300 minutes is far more likely to be the exception than the rule.
Click here to see the last three seasons of Kane compared to Patrick Hornqvist.
A lot of this comes from human's inability to inherently notice percentages. Evander Kane takes far more shot attempts per minute and in total than any other Jet. More total shots generally means a greater total in both good and bad shots. Humans notice the total not the percentages.
It's the same reason why fans tend to notice Dustin Byfuglien (as a defenceman) and Tobias Enstrom lose puck possession the most on the Jets, but do not notice that they actually lose possession per puck touch the least (Mark Stuart does the most often).
It's the same reason why fans tend to notice Evander Kane misses the net the most on the Jets, but do not notice that he actually misses per attempt near the Jets average (Wheeler and Scheifele miss more often per attempt).
Final thoughts on regression
Sometimes it is difficult to apply the concept of regression to real life. Regression is not bad luck making up for past good luck or vice versa.
A simple way to think of it is this: It is very possible to get 75% heads when you flip four coins; it is even highly probable. If you flip 1000 coins, the chance in having 75% heads is far less likely.
Coin-tosses are random -so it's not 100% analogous- but it gets the concept through. While Kane's play may have an impact on his percentages, over the long run he won't sustain those low numbers and they will move towards his career norms. If Kane does eventually start to dramatically shift in performance, his shot volume numbers will dramatically shift as well... not just the highly volatile percentages.
Evander Kane is not a perfect player. This article is not constructed to defend him of what he does. It does serve its purpose in showing how better to evaluate a player when analyzing their point production.