Does Corsi fail with Alexander Burmistrov?

After a two-season hiatus in the Russian KHL when Alexander Burmistrov exercised his legal right as a free agent, the young centre has returned to the Winnipeg Jets for the 2015-2016 and 2016-2017 seasons.

At an average salary value of 1.55 million dollars, this was generally thought of as a positive move. Burmistrov was highly regarded for his defensive acumen at a young age and highly effective penalty killing

Burmistrov was also fairly well regarded in his ability to improve shot-metrics and post a strong Corsi rating; however, he struggled to score points.

Does this struggle mean that Corsi over values Burmistrov?

This is not an unusual question to be seen asked. It has been seen speculated on Twitter.

In the Winnipeg Free Press’ preview to the Jets 2015-2016 training camp, Ed Tait and Tim Campbell wrote: “Many also touted (Burmistrov’s) strong possession numbers, but those were often a mirage given his perimeter habits.”

Were they really a mirage?

Why Care About Corsi?

goals

First off, let’s have a quick review on why one should care about Corsi.

Shot attempt differentials, or Corsi, generally should be preferred over goal differentials at smaller samples. The reason behind this is goal differentials are rare and so they do not predict future goal differentials very well.

In the short run, goal differentials regress towards Corsi. In other words: one caries about Corsi because they care about future goals more than past goals.

Goal differentials are highly volatile and need large samples to resemble talent, and the plus/minus statistic has inherent issues that make it essentially obsolete. So, Corsi is preferred metric to these two more traditionally used numbers.

Now, Corsi is not perfect, and does not tell the whole story. Its usage was never intended to act as a holistic statistic like Wins or Goals Above Replacement.

Rather, Corsi was akin to On-Base Percentage in baseball. 

OBP cannot and does not discern the true value difference between a homerun, a double, or a walk, just like Corsi considers all shot attempts as equal. If a team theoretically was perfect in these numbers, they would always win (or at least go to shootout in hockey). If a team posted zero in these numbers, the best possible outcome would be a tie (a shootout in hockey or infinite innings in baseball).

The reason why these numbers mattered was that teams were undervaluing their impacts, not that they were everything. Moneyball was never about statistics for sake of statistics. Moneyball was about a poor team taking advantage of a market inefficiency.

Eric Tulsky, now of the Carolina Hurricanes, showed that teams generally overpaid percentage drivers and underpaid shot differential drivers.

Did Burmistrov outscore his opponents?

Burmistrov is not a case where goal differentials paint a player differently than Corsi.

Over the two seasons Burmistrov played for Winnipeg, the Jets controlled 52.5 per cent of shot attempts with Burmistrov on the ice, while they controlled 52.9 per cent of goals.

In both cases, the Jets were much better with Burmistrov on the ice versus on the bench. The Jets’ Corsi% was 3.20 points better with Burmistrov, while Goal% was 4.93 points better.

Usage was only a minor impact as well. According to the most recent research, the largest outside factor to a player’s results is the strength of their linemates and teammates. Zone deployment and line matching then comes at a distant second and third.

When looking at zone-start adjusted numbers, we see that Burmistrov’s linemates did 2.5 points better in Corsi% and 2.5 points better in Goal% with Burmistrov than away from him. Burmistrov was not carried by his linemates, but instead improved them.

Analytically you would then describe Burmistrov as a player who improves goal differentials, and his Corsi suggests that his push was more likely sustainable and not variance (luck) driven.

What about Burmistrov’s perimeter play?

sh

The common complaint on Burmistrov was that he held onto the puck too much, dangled, and kept to the perimeter where shots do not matter as much.

This is a legitimate concern when it comes to numbers like Corsi; Corsi does not take into account that different shots are higher probability goals than others. While history has shown the sustainable control over shot quality is minor enough that ignoring does not damn Corsi in the general sense, there still may be cases of discrepancy in the specific.

The area size of each hexagon above represents the commonality of Burmistrov shooting from that location. Burmistrov’s shot distribution still shows him generating most of his offense from the prime areas.

The graph on the right shows Burmistrov’s shooting percentage relative to the rest of the league. Burmistrov has scored more often per shot on the outside than league norm, which may be the source of some of the perception on the young centre.

We can dig further into the numbers using factors, such as shot location, to improve upon Corsi into WOI’s scoring chance and high-danger scoring chance differentials.

The Jets controlled 51.0 per cent of scoring chances with Burmistrov on the ice, which was 1.5 points better than with him on the bench. The Jets also controlled 51.8 per cent of scoring chances with Burmistrov, which was 3.0 points better than with him on the bench.

Burmistrov also ranked above Jets’ centres Bryan Little, Mark Scheifele, Nik Antropov, Olli Jokinen, and Jim Slater in these numbers over the same time period.

Why the disconnect?

burmi

The disconnect comes from the weakness of the eye-test.

Neither eye-test nor numbers are perfect, but each has its own strengths.

The eye-test is heavily influenced by heuristics. The human memory relies on on shortcuts as we cannot remember all events. Instead, our memory saves up the few moments we feel subconsciously is most important.

In hockey this predominately means remembering goals and the people we saw directly contribute to those goals. This is why the layman eye-test tends to heavily follow goals, assists, shooting percentages, on-ice save percentage, and PDO. 

However, players can indirectly improve a team’s chance to outscore the opponent and Burmistrov is a good example of this.

Corsi in many ways represents a general tilt in the ice in terms of shots and chances. The Jets do better in these things with Burmistrov on the ice. That alone is worth some weight.

When we look into zone transition statistics, we can see why as well with Burmistrov being adept in both leaving the defensive zone and gaining the offensive zone with control of the puck. Burmistov is also exceptional in terms of preventing opposition transition, especially with his tenacious forecheck.

The Goals Above Replacement chart shows Burmistrov’s value predominately rests in his ability to repress opponents shots. About 69 per cent of his total value lies in Shotrate Against GAR. While the eye-test can detect defensive play, it cannot see defensive impact and the two are not the same.

Defensive impact is essentially all the events that never occurred, being able to detect this is nearly impossible by the eye-test.

Now, Burmistrov was not a perfect player. We see that he struggled in faceoffs and finishing talent. The latter is primarily what many are focusing upon when they dismiss the Russian centre.

Regardless, Burmistrov was still a positive influence on the Jets on-ice results. His Total GAR per game and per minute sat in the top six for Jets’ forwards over those years. His numbers extrapolate to a 7.71 GAR per 82 game season, higher than either Adam Lowry’s or Michael Frolik’s numbers last season.

All numbers are courtesy of WAR-on-ice and stats.hockeyanalysis.