Corsi, shot quality, and the Toronto Maple Leafs

[hatterson]

Log yourself onto extra skater.com, and use their handy tool to sort the top 5 teams in fen wick for percentage. (FF%). Look at the top teams on the chart. Then mosie on over to tsn.ca, and look at the overall league standings. Notice anything there? Perhaps, some similarities?

Top 5 teams in FF% in all situations are all in the top 6 for points overall. Now that can't just be irony can it?

First, irony has nothing to do with it. I think you'd be looking for coincidence.

Second, no one has ever claimed that Corsi, or Fenwick or Fenwick close or whatever variation you like doesn't correlate strongly with winning, the data clearly shows that it does, but that doesn't mean it's impossible for a team to maintain high shot quality for or low shot quality against and thus have results that aren't inline with the correlation.

Anaheim is 13th in FF% close but is 3rd overall in the standings. New Jersey is 6th in FF% close but is 21st in the standings.

It's like comparing wet grass to days when it rains. The correlation between the two is incredibly high, but that doesn't mean a guy can't water his lawn and have wet grass when it doesn't rain (or..uhh...dry his lawn? Yea this isn't the best analogy, deal with it)

 

[Micklebot]

I don't even have a horse in this race except the one named skepticism (and for good reason).

Shot quality is definitely a factor. Injuries are a factor. Goaltending performance is an obvious factor.

How long have people been chanting the regression was inevitable mantra for the Leafs? All of last season and up until 5 games ago when the Leafs go into a bad stretch with a depleted team and now it is time for the "I told you so".

You'll have to do better than predictions like that if you want people to take "advanced" stats seriously.

Is there a correlation between shot attempts and shots on goal and possession to winning? I'm sure there is.. is it an all determining factor? I'm sure it isn't.

To be fair, their ES sh% regressed quite a bit from last season, from 10.56% to 8.26% but it was made up for by their sv% jumping up from 92.39 to 94.47. In the end, the 2 cancel each other out, but I don't think it is unfair to think either were unsustainable individually. The problem as I see it is we're dealing with 2 seasons that combined only add up to 3579 mins of ES hockey or 76 games worth.

Toronto also has been fantastic this year on the PP, while being below average on the PK. A bit of a reversal of last year again, where they were fantastic on the PK and average on the PP.

So I guess what I'm saying is, they did regress, but made up for it in other areas. Last year they paced at 97 pts, this year they are on pace for 90. Not huge, but enough that it would have missed the playoffs if they played at that pace last year.

 

[danisonfire]

First, irony has nothing to do with it. I think you'd be looking for coincidence.

Second, no one has ever claimed that Corsi, or Fenwick or Fenwick close or whatever variation you like doesn't correlate strongly with winning, the data clearly shows that it does, but that doesn't mean it's impossible for a team to maintain high shot quality for or low shot quality against and thus have results that aren't inline with the correlation.

Anaheim is 13th in FF% close but is 3rd overall in the standings. New Jersey is 6th in FF% close but is 21st in the standings.

It's like comparing wet grass to days when it rains. The correlation between the two is incredibly high, but that doesn't mean a guy can't water his lawn and have wet grass when it doesn't rain (or..uhh...dry his lawn? Yea this isn't the best analogy, deal with it)

From the Article on fenwick I posted:

At the team level, Fenwick percentages are probably one the best predictors we have of future success. Teams that can sustain a high Fenwick percentage over the long term tend to be out playing their opposition at a high level. This also belies some teams that are succeeding due to unsustainable luck.

Take for example the first third of last season where the Minnesota Wild were leading the NHL standings, and the Los Angeles Kings were mired at the bottom of the Western Conference. Those Shot Reports from TimeOnIce.com cover the 2011-12 season from opening night until the end of Dec. 17th 2011. At that point last season, Minnesota had 45 points through 33 gp, with a 20-8-5 record and were leading the NHL. Los Angeles had a 14-14-4 record through 32 gp, and were ranked 11th in the West, 5 points out of a playoff spot.

At that point in the season though, Los Angeles had a Fenwick Percentage of 0.513, while Minnesota had a Fenwick Percentage of 0.426. This means LA was getting approximately 51.3% of the scoring chances at Even Strength in their games to that point in the season while Minnesota was only getting 42.6%.

Obviously there's a question around sustainability here. Minnesota went on to finish the year with 81 points in the standings, and a 35-36-11 record. Their Even Strength Close Fenwick Percentage to close out the year was 44.9%, the worst in the NHL. Los Angeles finished in 8th in the Western Conference with 95 points and a 40-27-15 record (digging out of an early hole can be hard to do), but they eventually went on to win the Stanley Cup (as most of you hopefully are aware). L.A.'s Close Fenwick Percentage at the end of the season was 53.6%, 4th best in the NHL.

 

[Michael Gary Scott]

First, irony has nothing to do with it. I think you'd be looking for coincidence.

Second, no one has ever claimed that Corsi, or Fenwick or Fenwick close or whatever variation you like doesn't correlate strongly with winning, the data clearly shows that it does, but that doesn't mean it's impossible for a team to maintain high shot quality for or low shot quality against and thus have results that aren't inline with the correlation.

Anaheim is 13th in FF% close but is 3rd overall in the standings. New Jersey is 6th in FF% close but is 21st in the standings.

It's like comparing wet grass to days when it rains. The correlation between the two is incredibly high, but that doesn't mean a guy can't water his lawn and have wet grass when it doesn't rain (or..uhh...dry his lawn? Yea this isn't the best analogy, deal with it)

Of course theres outliers, I mean the Leafs alone are an outlier. I think that I might have jumped the gun a bit with what I said, kind of made it out to having FF% being an end all stat which its not.

I mean, the answers for why NJ and Anaheim are outliers are found in other stats. NJ has a high FF%, but theyre having one hell of a time trying to score goals. Thus the reason for their low ranking in the standings. They can have the puck all they want, and theyve got some strong possession players on their team, but those same players cant put the puck in the net.

 

[habsfanatics*]

I suppose the rash of injuries have nothing to do with it, then..

Injuries definitely play a role here, but they weren't playing good before the injuries. If you thought a .957 SV % on the PK was sustainable, then I don't know what to tell you. When it came back to earth, they dropped from 2nd to terrible. I don't think the injuries are the biggest problem here. They're just not a very good team who was outscoring their problems for longer they probably should have.

 

[hatterson]

From the Article on fenwick I posted:

At the team level, Fenwick percentages are probably one the best predictors we have of future success. Teams that can sustain a high Fenwick percentage over the long term tend to be out playing their opposition at a high level. This also belies some teams that are succeeding due to unsustainable luck.

Take for example the first third of last season where the Minnesota Wild were leading the NHL standings, and the Los Angeles Kings were mired at the bottom of the Western Conference. Those Shot Reports from TimeOnIce.com cover the 2011-12 season from opening night until the end of Dec. 17th 2011. At that point last season, Minnesota had 45 points through 33 gp, with a 20-8-5 record and were leading the NHL. Los Angeles had a 14-14-4 record through 32 gp, and were ranked 11th in the West, 5 points out of a playoff spot.

At that point in the season though, Los Angeles had a Fenwick Percentage of 0.513, while Minnesota had a Fenwick Percentage of 0.426. This means LA was getting approximately 51.3% of the scoring chances at Even Strength in their games to that point in the season while Minnesota was only getting 42.6%.

Obviously there's a question around sustainability here. Minnesota went on to finish the year with 81 points in the standings, and a 35-36-11 record. Their Even Strength Close Fenwick Percentage to close out the year was 44.9%, the worst in the NHL. Los Angeles finished in 8th in the Western Conference with 95 points and a 40-27-15 record (digging out of an early hole can be hard to do), but they eventually went on to win the Stanley Cup (as most of you hopefully are aware). L.A.'s Close Fenwick Percentage at the end of the season was 53.6%, 4th best in the NHL.

I'm well aware of the strong predictive power of Fenwick. I'm also well aware that it's not perfect and it's far from proven that it absolutely can't be defied in the long run.

Hopefully things like the Shot Quality Project at Sportsnet will give us some insight into this.

 

[BraveCanadian]

Geesh.. outshot 50-24 last night and pull off a win.. crazy.

The Leafs sure are getting some mileage out of their goaltending duo so far this year.

 

[hatterson]

Geesh.. outshot 50-24 last night and pull off a win.. crazy.

The Leafs sure are getting some mileage out of their goaltending duo so far this year.

From a recollection of the game the chances were a lot more even. Granted that's from a biased and incomplete recollection of the game, but it would be interesting to see all of the corsi events mapped out.

 

[ark]

Geesh.. outshot 50-24 last night and pull off a win.. crazy.

The Leafs sure are getting some mileage out of their goaltending duo so far this year.

It's the story of our whole season this far. If either goalies, or worse, both get hit with the injury bug thats plaguing our team, I really do not want to know how our AHL goalies will be like night after night with 40+ shots.

 

[Michael Gary Scott]

From a recollection of the game the chances were a lot more even. Granted that's from a biased and incomplete recollection of the game, but it would be interesting to see all of the corsi events mapped out.

Heres your Fenwick chart from last night.

 

[hatterson]

Heres your Fenwick chart from last night.

Sorry I meant from where they were taken.

 

[matnor]

I haven't read through the whole thread, but can anyone direct me to a source which shows that shots on goals is a better predictor for future success than shooting percentage?

 

[The Mighty Duck Man]

Something that does trouble me about the Leafs isn't just the Corsi or Fenwick differences, but how many of those attempts they're allowing on goal. You look at another collapsing team like the Ducks, who also has been defying Corsi, and while they allow a lot of attempts toward the net, not many get through. I'm too lazy to go through every team, but quickly scanning, their SA/CA rate of 48.9 is the lowest in the league, while Toronto's 52.6 isn't good at all.

 

[GuineaPig]

I haven't read through the whole thread, but can anyone direct me to a source which shows that shots on goals is a better predictor for future success than shooting percentage?

Not exactly what you're looking for, but this article goes through a couple different stats and their predictive ability.

 

[Freudian]

It looks like they're plummeting down the standings. It's inevitable. You can't give up ten more shots a game on average and still compensate with good goaltending, good special teams and sharp counter attacking over a full season. The ice is too tilted for that to be possible.

Of course one wouldn't need corsi/fenwick to see this. The good old eye test is enough to establish this.

 

[BraveCanadian]

It looks like they're plummeting down the standings. It's inevitable. You can't give up ten more shots a game on average and still compensate with good goaltending, good special teams and sharp counter attacking over a full season. The ice is too tilted for that to be possible.

Of course one wouldn't need corsi/fenwick to see this. The good old eye test is enough to establish this.

I agree they have been playing poorly, but as of two games ago their record was the same as last season at the same point.

It will be interesting to see where they end up this year.

 

[doulos]

It looks like they're plummeting down the standings. It's inevitable. You can't give up ten more shots a game on average and still compensate with good goaltending, good special teams and sharp counter attacking over a full season. The ice is too tilted for that to be possible.

Of course one wouldn't need corsi/fenwick to see this. The good old eye test is enough to establish this.

Stats Geek - early season: "If you look at some of the underlying stats then it sure looks like the Leafs are playing way above their head. It seems reasonable that they will regress this season."

Saw Him Good Guy - early season: "Do you even watch hockey? If you watch the Leafs then you can see that they play a style of hockey that gives up lots of shots but takes advantage of the chances they do get and they win hockey games. Sure, the goalies have been unreal, but goalies are part of the team too. You don't know what you're talking about."

Stats Geek - Now: "Yeah ... so sure looks like things are regressing as the season goes along. No surprise here."

Saw Him Good Guy - Now: "Do you even watch hockey? If you watch the Leafs then you can see that they have been playing poorly. Don't need any of your silly stats to predict that. You don't know what you're talking about."

 

[Bear of Bad News]

What's with the pejorative term "geek"?

 

[doulos]

Interesting. I never use that word in that way any more since geek seems to have shifted use in pop culture in many ways. Was not meant to be used pejoratively.

 

[Freudian]

Stats Geek - early season: "If you look at some of the underlying stats then it sure looks like the Leafs are playing way above their head. It seems reasonable that they will regress this season."

Saw Him Good Guy - early season: "Do you even watch hockey? If you watch the Leafs then you can see that they play a style of hockey that gives up lots of shots but takes advantage of the chances they do get and they win hockey games. Sure, the goalies have been unreal, but goalies are part of the team too. You don't know what you're talking about."

Stats Geek - Now: "Yeah ... so sure looks like things are regressing as the season goes along. No surprise here."

Saw Him Good Guy - Now: "Do you even watch hockey? If you watch the Leafs then you can see that they have been playing poorly. Don't need any of your silly stats to predict that. You don't know what you're talking about."

Using advanced stats to knock in open doors it a bit pointless. We were able to identify teams overachieving before we had corsi, you know.

If anything Leafs serve as a useful lesson in how long a stretch teams can have where they overachieve. Leafs are 11 games over .500 over their last 81 games despite being massively outshot. Perhaps that should be a lesson for those who like to predict teams inevitably failing after 20 games of a season because they have 47% FF 5-on-5?

 

[doulos]

So when is it appropriate to use advanced stats?

It's only a "open door" now because the team has regressed exactly as the stats guys said they would.

It seems rather convenient to say that "Well of course we all saw this coming even without the silly stats" now that the silly stats are starting to look right all along.

 

[BraveCanadian]

It's only a "open door" now because the team has regressed exactly as the stats guys said they would.

Based on?

 

[Freudian]

So when is it appropriate to use advanced stats?

It's only a "open door" now because the team has regressed exactly as the stats guys said they would.

It seems rather convenient to say that "Well of course we all saw this coming even without the silly stats" now that the silly stats are starting to look right all along.

Rather convenient? Do you seriously believe that the advanced stats crowd were the lone voices having doubts about Leafs and they now are triumphantly proven right?

Pretty much everyone with the exception of some Leafs faithful knew it's unsustainable to be outplayed night in and night out and rely on goaltenders to get wins. Predicting them would lose more than they did initially is as bold as predicting Buffalo will have a rough year.

 

[Bear of Bad News]

Rather convenient? Do you seriously believe that the advanced stats crowd were the lone voices having doubts about Leafs and they now are triumphantly proven right?

Who are you defining as "the advanced stats crowd"?

Most of us view CORSI as an indicator, just like many other things. And none of us view stats as a replacement for watching games.

 

[doulos]

Rather convenient? Do you seriously believe that the advanced stats crowd were the lone voices having doubts about Leafs and they now are triumphantly proven right?

Pretty much everyone with the exception of some Leafs faithful knew it's unsustainable to be outplayed night in and night out and rely on goaltenders to get wins. Predicting them would lose more than they did initially is as bold as predicting Buffalo will have a rough year.

Again, rather easy to state that 'Pretty much everyone with the exception of some Leafs faithful' after the fact. That hasn't been my experience, but perhaps you're right.

In the end I actually agree with you, that people who watched the games should also have seen what was coming with this team. It's just that there sure were a lot of vocal folks who didn't. And it's nice to see them eat a little crow.