Tuesday, April 8, 2014

The Corsi Hockey League

There has been a lot of discussion over the past few years on the so-called possession stats: what they do/don’t measure, how they can be applied, whether they are causal or just correlated, etc.

There has also been some higher scope disagreement as to whether or not they even have relevance to the game of hockey: do they matter when it comes to the most important aspect from a fan’s POV, namely winning?

I hope to help answer this last question by showing some very basic results and counts of recent history. I have taken the season results of the past 6 years to see what the outcomes are. It should be said that none of these results can be taken as statistically gospel, as there simply aren't enough data points to come to a firm conclusion. However, I am going to display the facts, and you can draw your own opinion on this matter.

[Aside: “Corsi” is the metric of choice in the following examples. “CF%” is the ratio of a team’s Corsi-For/Corsi-Against. All of these stats have been taken from 5v5-close play as per http://stats.hockeyanalysis.com/teamstats.php.
Rank is determined by sorting each team by CF% in descending fashion. Therefore the best CF% team would be ranked 1st, and the worst would be ranked 30th.]

Starting Point
Playoff Teams = 96 (16*6)
Conference Finalists = 24 (4*6)
Stanley Cup Finalists = 12 (2*6)
Stanley Cup Winners = 6 (1*6)

What do good teams look like?

Average CF% of PO Teams = 51.55%
Average CF% of Conf. Finalists = 52.30%
Average CF% of SC Finalists = 53.39%
Average CF% of SC Winners = 54.48%
(NHL Average = 49.99%)

So it looks like being a CF> 50% team is good. Playoff teams have averaged above 51%, and SC Finalist teams have averaged above 53%.

Average CF Rank of PO Teams = 11.22
Average CF Rank of Conf. Finalists = 10.17
Average CF Rank of SC Finalists = 8
Average CF Rank of SC Winners = 5.83
(NHL Average = 15.5)

This seems to follow with the ranking results. An average of 11.22 means playoff teams are typically higher in the CF% rankings. Specifically, it looks like sitting in the top-10 of the league is where you want to find yourself; SC Finalists and Winners all seem to cluster near the top.

There seems to be a specific trend between postseason success and CF% and CF Ranking, noticeably above the normal NHL averages.

* * * * * * * *

But this perspective is only one side of the argument (Playoffs v CF%). Perhaps we should look from the other angle (CF% v Playoffs)…

How often does CF50 come up?

  • PO Teams: 74 playoff teams were CF>50%
  • 74/96 = 77.08%
  • Conf. Finalists: 17 conference finalists were CF>50%
  • 17/24 = 70.83%
  • SC Finalists: 10 Stanley Cup finalists were CF>50%
  • 10/12 = 83.33%
  • SC Winners: 5 Stanley Cup winners were CF>50%
  • 5/6 = 83.33%

So there were a very high number of good Corsi teams where it matters – in the playoffs, in the Conference & Stanley Cup finals, and as Stanley Cup champions. The proportions have many more CF>50% teams succeeding (74, 17, 10, 5) than CF<50 (22, 7, 2, 1, respectively)

What does CF50 get you?
In the 6 years I looked at, there were 96 teams that had a season of CF>50%. 74 of those teams made the playoffs.
74/96 = 77.08%
So if you have a team Corsi above 50%, your chances of making the playoffs are over 3/4.

Conversely, there were 84 teams with a season of CF<50%. Only 22 of those teams made the playoffs.
22/84 = 26.19%
Being a sub-50% team doesn't look as positive. Being <50% meant you made the playoffs just over 1/4 of the time.

Do you want to be a top-ranked CF team?
When you compare the Top-16 ranked CF% teams vs the CF50 results, they happen to line up exactly with the figures above: Only 5 times was a team CF50 and not finish in the Top-16, but 3 of those teams still made the playoffs; while 5 teams had a sub-50% CF rating and still finished in the Top-16, and 3 of those teams did make the playoffs.

Therefore, if you are a Top-16 team in CF%, then you've made the playoffs 77% of the time (74 out of the 96 Top-16 teams).
And as above, if you were a Bottom-14 team, then you made the playoffs only 26% of the time (22 out of the 84 Bottom-14 teams).

So being high in the Team CF rankings looks like a good thing.

* * * * * * * *

Food for thought
Best average CF% over the 6 seasons I looked at:
1.  Detroit: 55.42% (6 playoff appearances, 10 series wins, 1 cup)
2.  Chicago: 54.02% (5, 10, 2)
3.  San Jose: 53.37% (6, 6, 0)
4.  New Jersey: 52.43% (4, 3, 0)
5.  Boston: 52.4% (6, 9, 1)

Worst average CF% over the past 6 years:
26. Anaheim: 47.93% (4, 1, 0)
27. Colorado: 47.38% (2, 1, 0)
27. Winnipeg: 47.38% (0, 0, 0)
29. Minnesota: 46.48% (2, 0, 0)
30. Edmonton: 45.2% (0, 0, 0)

* * * * * * * *

Where does this leave us?

Well, that depends on your openness to the results of data. I think I've shown that the recent past shows a connection between possession stats (here, 5v5 close Corsi) and season results:
1) Successful teams (playoffs, conf. finals, etc) all seem to show higher Team CF%. 
2) And in the reverse, the high CF% teams have a much higher proportion of success than the low CF% teams.

My opinion is that it’s much better to be a good CF% team than a bad one. This doesn't mean that you can’t have success when you are a bottom-level CF team, it just means that the likelihood of your success is much lower than if you were a top team.

If I was building a team, I would try and get my team to be in the top end of CF%. This would suggest a longer period of sustained success, rather than going against the grain and hoping to stumble on some magic formula that tries to beat the odds.
And if my team had a history of terrible CF%, I would treat it as a major symptom and look everywhere to find a cure, whether in the players, the systems, or the coaches.

No comments:

Post a Comment