A Layman’s Attempt at Speaking Corsi with ‘everyman’

This post will mark the first time I use my new found “influence” to really push a personal agenda. This is an attempt at settling an old score with a group of people hell bent on pushing a wholly fictitious concept called Jorsi. Jorsi was born out of one man’s attempt to discredit Corsi and my support of it. It is a byproduct of years spent toiling the roadways of Nisku’s Industial Business Park, finding common ground among the ruffian and hard working people of the Alberta oilfield industry with the goal of selling what the aforementioned need…

Actually that is all BS. It’s an old friend’s attempt at getting me all hot and bothered about advanced stats through making up some stupid advanced stat that means absolutely nothing BECAUSE CHARACTER, GRIT AND INTANGIBLES are all that really matter. BOY DOES IT WORK

Jorsi is defined as the Losing team’s higher Corsi+Fenwick minus Winning team’s lower Corsi+Fenwick = Jorsi rating #characterwins #defylogic. You can take it for what it is worth which is unintelligible gibberish. It is actually pretty funny if you can wipe away the history of confrontation between the warring factions of me against the world.

Anyways, the topic has taken one quite the negative tone among this group of friends and I. We’ve narrowed the discussion to little passive aggressive digs on social media and I, like them I’m sure, am growing tired of it. They send a volley of Jorsi success, I tell them to dive over Niagara Falls. They say how stupid Corsi is. I tell them that they are ignorant asses and they need to just open their minds a little and stop thinking like old men. It’s PLENTY of fun.

Lets put it all on the table! Me against you! Corsi and Advanced Stats VS the nonsensical & imaginary Jorsi.

I’m going to start the justification of Corsi with a small statement. I am not an expert. All that I know is through my own deduction and understanding of what I read and follow. There are undoubtedly mistakes in my use of Corsi and other metrics, but what I know is that there are relatively easy ways to prove if there are some correlations between good teams and good CF%. And that is at the crux of this never ending circle of jabs. “They” think it is stupid, and I see merit in it.

Today there was, I believe, a small breakthrough.

Here goes nothing!

First off, for those of you who are new to advanced stats, who’s eyes glaze over at first sight or who’s ears pour steam at the very mention of the words advanced and blogger, I would like to give you a short explanation of what Corsi is. (drum roll) It basically is…

• The number of shots attempted, whether they make it to the net or not.
• There are Corsi events for and Corsi events against.
• CF% is the ratio, in percent, of events for to events against.
• For example, if a team takes 50 attempts and allows 50 it has a CF% of 50%. If a team takes 35 attempts and allows 45 it has a CF% of 43.75% or what can be called a typical Oilers CF%.

Ok. If you need, take a minute and let that sink in.

Got it? Good. Lets move on.

In hockey the number of shots a team takes and gives up matters. Corsi is just an extension of shots. Not all attempts or shots are made equal, but Corsi is a decent representation of who has the puck more. If you have more shots and attempts chances are your team has the puck more and is better at getting the puck back when lost. The other team cannot score if they don’t have the puck. POSSESSION.

Now that we have all that out of the way I would like to get back to the point of this post. Ending the reign of Jorsi. As mentioned, I believe that I may have made a breakthrough. This break happened as result of one of those passive aggressive social media jabs. The jab went as follows.

Any idea when the Avalanche are supposed to start losing???

The tweet was one in a short string of tweets condemning Corsi. Of course Corsi is useless! The Avs haven’t started losing yet even though the Corsi guys said they would. BAH HA HA! Oh hot mama it got me going. I promptly sent out my usual niceties in response. But this fairly innocuous tweet got me thinking. How do I end this viscous cycle of statitude vs grititude? An email was crafted several days later.

Hey,

Since your “Any idea when the Avalanche are supposed to start losing???” tweet, I’ve been thinking a lot about Corsi and advanced stats. I’m going to be making a post about this in the next few days but I still wanted to point out a few things to you before hand. (all the numbers include every game situation)

1. As of today, Colorado sits in 8th in the West.
2. They are the 11th best team in the west according to Corsi and have a CorsiFor% of 47.2%
3. Their last 3 games they have put up a 6.13 sh% and .856 SV% and went 0-3
4. The first 16 games they recorded an 11.28 SH% and .941 SV% and went 14-2
5. On the season they are also sporting the #1 SV% at .932
6. Currently they are recording the 3rd best SH% in the west at 10.2%
7. They have given up the 8th least shots.
8. They’ve taken the 12th most shots in the west.
9. Their SHF% is 49.0% which is the 10th best in the west.

The “experts” have been saying this will not last for Colorado because in all likelihood (a statement addressing the probability, not the definitive outcome), the numbers they have been achieving won’t last. Their save percentage was out of this world for the first part of the season, it won’t (and didn’t) last. In conjunction to that, they have been putting up an unsustainable SH%. It probably won’t last, although there are outliers season to season (see Toronto last year). To top it off, they are an average possession team (shown with Corsi). Shots for and against matter. Corsi is an extension of shots for and against.

What is the point of this you ask?I find it incredibly irritating when someone like yourself (a relatively intelligent individual IMO) takes the angle of the majority, which is in large part built on uniformed, opinionated assumptions. Your question, mocking some pretty proven numbers, is something I expect of a lesser person. The Corsi and advanced stats community are trying to figure out new ways to analyze hockey through standardized means and they have made huge strides in the past few years in doing so. Nothing is 100%, including both stats and someone’s view/opinion based on experience(s). I see the applications, understand the numbers they have developed and find them incredibly compelling and fascinating.

Colorado may make the playoffs. Their 14-2 start gives them a chance. They are an improving possession team according to Corsi & shot numbers and if they can maintain a break even Corsi, shotsfor% (both have hit the break even point at about the 10-14 game rolling mark and generally stayed there), get average goal tending and good luck in the SH% department (shooting percentage is very unpredictable) they may have a chance. But I for one will not be at all surprised if they don’t make it to the playoffs because the west is so deep and competitive.

Hockey is still played on the ice and tactics, motivational abilities, emotion, schedule, injuries, experience level, skill, size, etc, all still play a very large part in the outcome of a game. In individual games they have more effect, but over the long term things even out and the narratives about the intangibles becomes grossly over-hyped. Numbers, especially over large samples, don’t lie. Very high numbers in SV% and SH% are usually due to a short term string of good luck. Conversely it goes for unusually low numbers. They are a predictor of future outcomes and the more work that is done will lead to even more prediction and analysis power. They are not perfect and will never be 100%, as hockey still depends on the above non-quantifiable variables, but neither will a persons judgement provide 100% certain outcome or a coaches tactics or skills of any player(s).

Here are some graphs for you to look at.

corsi for colorado charts.xls

PLEASE take some time to look at them.

All the numbers I’ve used come from www.extraskater.com (like I said above, the numbers are based on minutes in all situations, even strength has proven to be the best predictor of future success though)

A few things I want you to keep in mind

• In my opinion (based on what I have read and followed) Shot rates and Corsi rates (and all the derivatives and common numbers) are a result of tactics and ability of players to execute those tactics.
• Shooting percentage is not easily controlled and has a large amount of luck, relative to the skill level of the players shooting. Really high %s are almost all luck, team wise or individual player wise. It almost always trends a back to average and the longer you go the more likely that regression is (positive or negative).
• SV% is also has a segment of luck but tends to be a little more consistent. But when goalies are putting up numbers in the .930 and up range it will, with almost 100% certainty, not last more than a handful of games. There are exceptions from season to season, but they are not the rule.
• When you add together high SH% and High SV% you can guarantee that the underlying numbers won’t reflect what is going on. That is why even though Colorado’s Corsi is below average, and there are many games when they were/are out Corsi’d and out shot, their lucky SV% and SH% have kept them winning and at or near the top to this point. Now that their numbers are trending to average (sv% and Sh% down, corsi and shots up) their record and performance, over larger segments will be far more accurate of how good or bad a team they really are.
• The whole point of the numbers is to be able to see through luck and start to see teams for what they really are.

If you want to talk about this lets talk. I don’t want to argue about this anymore in the fashion that I do with you and the other guys.

Hey Boris,

I appreciate the email, both for effort and honesty. I think the best way to formulate my response is to address it in two parts:

• Corsi – I find it admirable that you have researched this topic to this extent. You are obviously passionate about this type of analysis, and the theories behind it. The introduction of your blog and podcasts also tell me this is something that drives you, and you have found a genuine interest in it. I say thats fantastic. To explain my stance, I will say it’s partially based on my belief that hockey is the type of game where finite and specific stats are not accurate enough to be used as metrics to judge success or failure. It’s a game with too much flow (unlike baseball or football) and therefore intangibles should and do play a siginificant role in outcome (shot quality, clutch scoring, big saves, 1 on 1 battles, turnovers in crucial areas, etc). In saying that, and this does support what you are saying, my stance against Corsi is due to ignorance. I admit it. Although I have accumulated some knowledge, the extent to which it affects players, seasons, and trends, like you have described below, is something I have not researched. Nor do I care to. That is not intended as a “jab” either. I just enjoy watching hockey for hockey, for the game it is, without looking between all the lines and breaking it down to this extent. Dont get me wrong, I like stats and think they are an important part of any game, but will probably never disect a team or season such as you have done. I respect the time you have devoted to this, and I say continue…

THERE IT IS FOLKS! Two adversarial factions finding understanding in each others views. I’m am OK with his reasoning for not buying in. I get it. It takes time, energy and desire to start to get a grasp on all the numbers, but that is not to say that it is not possible. I’m a firm believer that almost anyone can be taught a concept as long as that concept is delivered in a way that the person learning can understand it.