3 of the top 4 Corsi teams in the league are in the Final Four.

[dgibb10]

You really need to pause and research statistics (gain a better understanding) before commenting so confidently. This isn't the first time. You do drive-bys with blatantly incorrect data all the time. It's starting to almost look intentional at this point.

Like, in this very thread you are advocating for CORSI (which is SHOT ATTEMPTS) including special teams and non 5v5 play....I guess the good thing is all of us disagreeing in this thread on other things can probably unite on this issue.

This is incoherent.

Those were the top 12 teams in corsi (CF%) this year. 9 made the playoffs.

 

[StrangeVision]

Corsi strikes me as a descriptive rather than predictive metric. Is a team who plays to maximize their corsi going to have more wins or is it just that better teams happen to have more possession and take more shots? I lean towards the latter.

Consequently, a team that has the ability to maximize their corsi is going to be a better team, because you need possession to take shots and you need to be good to have more possession, but it's not because they're trying to maximize their corsi. Trying to have a good corsi isn't the goal, a good corsi is simply a byproduct of being a good team.

 

[dgibb10]

Corsi strikes me as a descriptive rather than predictive metric. Is a team who plays to maximize their corsi going to have more wins or is it just that better teams happen to have more possession and take more shots? I lean towards the latter.

Consequently, a team that has the ability to maximize their corsi is going to be a better team, because you need possession to take shots and you need to be good to have more possession, but it's not because they're trying to maximize their corsi. Trying to have a good corsi isn't the goal, a good corsi is simply a byproduct of being a good team.

You just described a predictive metric.

Shot attempts lead to shots which leads to goals. Goals being the descriptive metric (awards should be based on descriptive metrics), and the shot attempts being the predictor.

 

[StrangeVision]

You just described a predictive metric.

Shot attempts lead to shots which leads to goals. Goals being the descriptive metric (awards should be based on descriptive metrics), and the shot attempts being the predictor.

I guess my point was if you take an untalented team and tell them to attempt more shots to get their corsi up, it's not going to lead to more wins. You still need to actually be good to win games, you need talented players who can put the puck in the net. Getting your corsi up isn't necessarily going to lead to a better record, it's simply a byproduct of already being a good team.

 

[dgibb10]

I guess my point was if you take an untalented team and tell them to attempt more shots to get their corsi up, it's not going to lead to more wins. You still need to actually be good to win games, you need talented players who can put the puck in the net. Getting your corsi up isn't necessarily going to lead to a better record, it's simply a byproduct of already being a good team.

But the argument isn't at an individual player level tho or a coaching level telling players to simply "generate more corsi". Since you cannot simply just make more shots out of thin air (and if you do those poor quality shots would be recognized with more advanced metrics like expected goals).

It's about from a roster building perspective. Finding guys who dominate possession and generate lots of shot attempts naturally, as well as preventing them the other way.

 

[StrangeVision]

But the argument isn't at an individual player level tho or a coaching level telling players to simply "generate more corsi". Since you cannot simply just make more shots out of thin air (and if you do those poor quality shots would be recognized with more advanced metrics like expected goals).

It's about from a roster building perspective. Finding guys who dominate possession and generate lots of shot attempts naturally, as well as preventing them the other way.

So you agree it's a byproduct of being good, great.

 

[dgibb10]

So you agree it's a byproduct of being good, great.

Can you name a stat that isn't a "byproduct of being good" by your logic?

 

[StrangeVision]

Can you name a stat that isn't a "byproduct of being good" by your logic?

I mean, better players/teams will have better stats, that's a simple fact of the life. The question here is what causality is there between corsi and goals/winning? If a team decides to make a concerted effort to increase their corsi, can they expect their goals/wins to go up? I have my doubts, if the talent level doesn't change. Alternatively, if a team wanted to increase their high danger area scoring chances, I'd say that would have a better chance of increasing their goals than simply trying to take more shots from anywhere on the ice.

I just don't see the usefulness in pointing out that the best teams have the best corsi scores like it's some kind of validation of its predictive quality, it's basically just common sense that better teams would have a better corsi because it is a description of possession, they're not out there specifically trying to maximize their corsi.

A team could theoretically attempt a shot every moment they enter the offensive zone, it doesn't matter what kind of shot or where it ends up or where on the ice you took it, and their corsi would improve. Does that actually help them win, though? That's where corsi loses its value to me as a predictive metric. It's not something a team can use to improve their chances of winning, it's a description of how good a team is at possession.

A good corsi doesn't cause a team to be good, it is just correlational.

 

[FiveFourteenSixOne]

When woodcroft was fired didn't they notably have strong metrics? Just lack of scoring with McDavid hurt and goaltending woes.

Yes they did.

I'm pretty certain that if they rode it out without firing Woody their season would have turned around regardless of the coaching change. It may not have improved as much as it did, but they easily would have made the playoffs.

They never would have had the playoff success they did with Woodcroft, though.

Knob is significantly better at making tactical adjustments than Woody ever was, whose only strategy was moving McDrai on to or off of the same line.

 

[daver]

Isn't that what Marner yelled?

CORSI THE F*** UP!!!

 

[Oak]

One of these teams isn't like the others:
View attachment 1041194

Seems like there is another metric we should be judging team quality by.

 

[PenguinSuitedUp]

Did next to nothing to improve a team he inherited.

He’s been GM 1 year and essentially traded Necas, who didn’t want to play in Carolina, for two 1st round picks and Stankoven. Looks good so far for me.

 

[JaegerDice]

Paging OP @JaegerDice since you are replying to folks. Enjoy the light reading above and defend your misrepresentation in the title and OP. As a reminder, you claim CORSI by itself is very predictive in winners and top playoff teams.

Uh, if you could quote where I claim that, that would be great.

Let me save you the time. I never said that. I stated a series of simple, proveable facts about the final four this season and the final last season, and then opined "Crazy how predictive this stat continues to be."

Corsi does not single-handedly predict winners. Nobody, including I, has ever claimed it as a 1:1 correlation with wins or playoff wins. It simply has a higher correlation to future 5v5 goal differential and future wins than any other single hockey metric, and most traditional hockey metrics put together.

Expected goals models that weigh a multitude of metrics including Corsi, are more predictive than Corsi in of itself. Advanced hockey stats are continuing to develop as pretty much every NHL team pours millions into analytics departments, hire GMs and AGMs out of analytics departments, buy out and hire the creators of third party analytic sites and models. And of course, there are still independent minds out there looking at hockey in different ways, building their own models, bringing new ideas to the table, disrupting and driving things forward.

That doesn't change my opinion that it's crazy (or rather, impressive) how predictive Corsi continues to be compared to the eye test or other traditional box score measures, let alone narrative nonsense like 'durrr grit!' or 'the will to win!'.

Like being impressed at the seed that lead to the tree.

 

[Edmonton East]

Last season the 2nd best and 3rd best teams met in the Final.

If there's a rematch this year, it will again be a meeting between the 2nd best and 3rd best team.

Crazy how predictive this stat continues to be.

The thread title, this OP, and your subsequent posts in this thread must be your alter ego then I guess. Let me know when he/she wants to refute any of my previous points.

There you go again in the last post trying to claim Corsi is some big brain statistic that strongly predicts results and is vastly better than the silly eye test and actually watching the games. Again, 42% of the top 12 corsi teams missed the playoffs. So let's try this since previous examples and data don't seem to be getting through yet:

Do you honestly believe most people that watch and follow hockey religiously couldn't predict the top 12 contenders with better accuracy than Corsi? Who that watches hockey would say the Preds (11), Pens (10), Utah (6), Flames (7), or my Devs (8....play fell off pre injuries btw)...are top teams in this league? Or even above average? Who would say those teams are better than Corsi #29 Toronto or #18 Dallas? Who would say the Canes are completing their dynasty this year since they've far and away dominated corsi 3 straight years?

 

[JaegerDice]

The thread title, this OP, and your subsequent posts in this thread must be your alter ego then I guess. Let me know when he/she wants to refute any of my previous points.

There you go again in the last post trying to claim Corsi is some big brain statistic that strongly predicts results and is vastly better than the silly eye test and actually watching the games. Again, 42% of the top 12 corsi teams missed the playoffs. So let's try this since previous examples and data don't seem to be getting through yet:

Do you honestly believe most people that watch and follow hockey religiously couldn't predict the top 12 contenders with better accuracy than Corsi? Who that watches hockey would say the Preds (11), Pens (10), Utah (6), Flames (7), or my Devs (8....play fell off pre injuries btw)...are top teams in this league? Or even above average? Who would say those teams are better than Corsi #29 Toronto or #18 Dallas? Who would say the Canes are completing their dynasty this year since they've far and away dominated corsi 3 straight years?

I will continue to wait for you to quote me saying that Corsi single-handedly predicts winners, or has a 1:1 correlation with success.

Until then, I don't know who you think you're winning over with this 'old man yells at clouds' schtick regarding analytics, but I'll leave you to it Grandpa Simpson.

 

[Fatass]

There’s a reason why every club has an analytics dept. now. These stats are useful as part of the whole picture. More information is good.

 

[Tasty Biscuits]

Good teams probably bypass as many opportunities to shoot than bad teams.

But good teams generate so many more opportunities they're gonna shoot more just by default.

 

[NotCommitted]

I just don't see the usefulness in pointing out that the best teams have the best corsi scores like it's some kind of validation of its predictive quality, it's basically just common sense that better teams would have a better corsi because it is a description of possession, they're not out there specifically trying to maximize their corsi.

Isn't "best teams" dictated by points in regular season though? Going with that, many of the best teams this season didn't have good corsi at all, Jets & Capitals were very much middle of the pack both 5v5 & all strengths, while Toronto was one of the worse teams in the league.

 

[Feanor]

I mean, better players/teams will have better stats, that's a simple fact of the life. The question here is what causality is there between corsi and goals/winning? If a team decides to make a concerted effort to increase their corsi, can they expect their goals/wins to go up? I have my doubts, if the talent level doesn't change. Alternatively, if a team wanted to increase their high danger area scoring chances, I'd say that would have a better chance of increasing their goals than simply trying to take more shots from anywhere on the ice.

I just don't see the usefulness in pointing out that the best teams have the best corsi scores like it's some kind of validation of its predictive quality, it's basically just common sense that better teams would have a better corsi because it is a description of possession, they're not out there specifically trying to maximize their corsi.

A team could theoretically attempt a shot every moment they enter the offensive zone, it doesn't matter what kind of shot or where it ends up or where on the ice you took it, and their corsi would improve. Does that actually help them win, though? That's where corsi loses its value to me as a predictive metric. It's not something a team can use to improve their chances of winning, it's a description of how good a team is at possession.

A good corsi doesn't cause a team to be good, it is just correlational.

The whole point of corsi has always been that it is the best approximation of puck possession that we can derive from the data that is easily available.

The predictive nature of corsi comes from the fact that puck possession tends to improve the odds that a team wins, since most of the time you have to possess the puck in order to score and the opponent cannot score if your team holds the puck.

If we would get actual and sufficiently precise « puck possession « data from the tracking technology, it would most likely outperform corsi as indicator of future success. And it is easy to see how the teams could improve their puck possession numbers and how that would contribute to their ability to win.

If a team would « game corsi » by constantly shooting from anywhere, the fact that would contribute towards their winning chances would not so much be the shots fired, but the fact that in order to maintain good corsi numbers they would have to relentlessly hound the puck after each shot in order to win it back.

 

[JaegerDice]

A team could theoretically attempt a shot every moment they enter the offensive zone, it doesn't matter what kind of shot or where it ends up or where on the ice you took it, and their corsi would improve. Does that actually help them win, though? That's where corsi loses its value to me as a predictive metric. It's not something a team can use to improve their chances of winning, it's a description of how good a team is at possession.

Putting aside the fact that no player or team plays like this, this is actually a good example of a misunderstanding that people have about Corsi.

"A team could theoretically attempt a shot every moment they enter the offensive zone, it doesn't matter what kind of shot or where it ends up or where on the ice you took it, and their corsi would improve."

The thing about Corsi is that the shot attempts themselves are not individually important. The shot attempt simply tells us that the team had the puck, because you need to have the puck to attempt a shot.

In a weird way, everything that happens BETWEEN the shot attempts is more important than the shot attempts themselves. That includes, chasing down the rebound, winning races, winning 50/50 puck battles, making passes, keeping the puck alive in the zone... all the little things that 'eye test' aficionados claim stats don't take into account, are all captured in Corsi, because they're all necessary for a team to take a shot, get the puck back, take a shot, get the puck back, etc.

Teams that take a shot into the shin-pads, and then allow the other team to take one or more shot attempts for themselves are NOT strong corsi teams. At BEST, they will trade one attempt for one attempt, leading to a 50% Corsi, which would be middle of the pack.

In a hypothetical world, yes, a team could 'boost' their corsi by just taking any shot ever, but that team would still have to do all the little things to get the puck back and keep shooting for their corsi to go up relative to their opponent.. and so long as there's one puck in hockey, as long as that hypothetical team is doing so, their opponent doesn't have the puck to even attempt a shot on net.

In reality, there is no system in hockey that tells players to take a worse quality shot when a better quality play or opportunity is available.

 

[Edmonton East]

I will continue to wait for you to quote me saying that Corsi single-handedly predicts winners, or has a 1:1 correlation with success.

Until then, I don't know who you think you're winning over with this 'old man yells at clouds' schtick regarding analytics, but I'll leave you to it Grandpa Simpson.

Never said 1 to 1. Cute attempt at a strawman tho bceause i know you can't argue the facts. I continue to ask why you presented the data in a misleading way. I provide several reasons your posts are misleading or disingenuous at best. You keep directly claiming Corsi provides vastly superior insights that those simpletons that just watch the game in their caves couldn't possibly understand. You keep alluding to Corsi alone having some crazy strong ability to predict top playoff teams. When confronted with historical facts that dispute this assertion, you simply ignore them LOL. Still waiting for how Corsi would better predict the top 12 playoff teams vs someone that just follows the game religiously. Take your time.

This is like me creating a thread that the top corsi team hasn't won the Cup in the last 10 years, man this stat sucks! Which would be just as stupid.

I do love the old man trope folks that are too ignorant to "understand what they don't understand" or admit they are wrong naturally default to. Saw some previous guy call me a marketing professional or something lol. Hint: If you've ever actually known or worked with good statisticians or solid folks in analytics, they tend to be the folks that get the most irritated when folks like yourself mislead others. It's been a fun ride from being an early champion of analytics in hockey (from over a decade ago) to coming full circle and having to constantly refute folks blindly parroting, cherry picking, and/or extrapolating "advanced" stats they don't understand...or those that condescendingly write off ThE eyE TesT.

TL;DR: people need to stop trying to apply the rigidness of moneyball analytics to hockey. Baseball is a static sport. Hockey is very obviously not. Lastly, and again, saying you generally prefer more shot atempts than your opponent isn't some profound statement. Duh. But Corsi very ironically is not more predictive of top playoff teams than the eye test you hate so much.

 

[JaegerDice]

You've failed to present a single compelling argument for why a simple list of true statements is 'misleading'.

I have never alluded to Corsi being the only stat with any predictive value, I have simply stated the truth that it has the highest correlation to future goal differential and future wins than any single other hockey statistic and many traditional statistics combined. It is also less predictive than Expected Goals, which weights Corsi along with several other stats, both traditional and advanced.

You can present as many 'gotcha' attempts as you like, the R-squared for Corsi to future goal differential and future wins is higher than any other single statistic, proven again and again as the sample of NHL seasons since on-ice events were tracked grows. This is why Corsi continues to be weighted highly in every expected goals model, public and private.

If you want to have a conversation about the value of inputs vs outputs, or descriptive vs predictive value, then I'm open to it, bu tantruming over an argument that's already been decided is pointless. The hockey analytics 'debate' is over. Analytics won. That's why damn near every team is investing millions into it, 31 out of 32 NHL teams pay Sportlogiq for their analytics reports (which feature Corsi prominently), and why NHL teams are buying up every major 3rd party analytics site they can.

If you wholly believe in the eye test, the same test that 'proved' the Earth was flat and the Sun rotated around the Earth, then god be with you.

Finally, the idea that analytics can explain baseball but not hockey is a limitation of your imagination, not of math or statistics. Hockey is not magic. Rest assured if math and statistics could put man into space, they can certainly be used to understand and quantify whacking a rubber puck into a net with a stick.

 

[Bank Shot]

I guess my point was if you take an untalented team and tell them to attempt more shots to get their corsi up, it's not going to lead to more wins. You still need to actually be good to win games, you need talented players who can put the puck in the net. Getting your corsi up isn't necessarily going to lead to a better record, it's simply a byproduct of already being a good team.

A bad team won't be able to get their corsi up because they lack the talent to achieve it.

You need to be good in order to do that. Stats like corsi are just another way to measure good play.

If you are an NHL GM and want to improve your team, the best way to do it is by acquiring players that will improve your 5v5 play. Possession players. Corsi players. Your Philip Danaults.

If a team improves their possession play, they will almost certainly get better results.

It's because 5v5 is the majority of the game (like 70% of goals). So the best teams are generally the best 5v5 teams.

 

[dgibb10]

I mean, better players/teams will have better stats, that's a simple fact of the life. The question here is what causality is there between corsi and goals/winning? If a team decides to make a concerted effort to increase their corsi, can they expect their goals/wins to go up? I have my doubts, if the talent level doesn't change. Alternatively, if a team wanted to increase their high danger area scoring chances, I'd say that would have a better chance of increasing their goals than simply trying to take more shots from anywhere on the ice.

I just don't see the usefulness in pointing out that the best teams have the best corsi scores like it's some kind of validation of its predictive quality, it's basically just common sense that better teams would have a better corsi because it is a description of possession, they're not out there specifically trying to maximize their corsi.

A team could theoretically attempt a shot every moment they enter the offensive zone, it doesn't matter what kind of shot or where it ends up or where on the ice you took it, and their corsi would improve. Does that actually help them win, though? That's where corsi loses its value to me as a predictive metric. It's not something a team can use to improve their chances of winning, it's a description of how good a team is at possession.

A good corsi doesn't cause a team to be good, it is just correlational.

This is a really dumb strawman against something that isn't real.

Sure if a team abandoned trying to play normal hockey for the sake of artificially boosting shot attempts, it wouldn't work. This, again, applies to every stat.
If a team never took a single shot unless it was a high danger shot, the same would happen.

And it would be noticed instantly and such a team/player could be immediately disregarded by any intelligent person as an outlier.

Corsi is a predictive metric in that it is a very good underlying indicator that a team or player (playing normal hockey) is winning possession and winning their minutes, that may be hidden in raw stats by either very bad or very good shooting luck.


The usefulness is that you can find undervalued teams or players if they are dominating possession and just getting bad luck. You can get on board there and when the luck evens out, you will come out ahead. Or the opposite, you can find overvalued teams or players.

Of course, you can notice outliers. Elite shooters will consistently outperform, and some guys who can't finish for shit will underperform. By looking at a sample size of a few years however, these outliers are pretty easy to identify.

 

[dgibb10]

A bad team won't be able to get their corsi up because they lack the talent to achieve it.

You need to be good in order to do that. Stats like corsi are just another way to measure good play.

If you are an NHL GM and want to improve your team, the best way to do it is by acquiring players that will improve your 5v5 play. Possession players. Corsi players. Your Philip Danaults.

If a team improves their possession play, they will almost certainly get better results.

It's because 5v5 is the majority of the game (like 70% of goals). So the best teams are generally the best 5v5 teams.

He's trying to argue against a strawman team that ignores anything resembling actual hockey and just repeatedly fires shot attempts from center ice.