Post-Game Talk: Holl wins it in OT. Leafs defeat the Sens 3-2

[Gary Nylund]

You know what's funny? This is the exact style of game that Andersen has allowed a 3rd goal against lately so we lose in regulation. Control the play, control the chances, have a relatively small play end up in our net and lose. Campbell basically put 2 pucks in his net last night and it was still enough

Not sure what your point is. If it's that Campbell played well last night and with Andersen perhaps he plays worse and we lose then sure, that could have happened.

If it's that the only issue with our play during the slump is goaltending, then no.

I'd say it's a meaningful reflection of the process, but not the result. You want the process to drive the result.

I do find if interesting how often I've been willing to build on stats for actual interpretation, but it's usually dismissed. If people don't want (or can't understand) the eye test, but also ignore stats, whats left aside from bias?

xGF is more of a team reflection, while GF/GA would reflect situations and goaltending a bit more.

I don't ignore stats, they can be helpful so why not use all the tools at our disposal. On the other hand, when people post Xgoals stats for one game (or even a series) as proof of something, for me personally, I believe the stat to be flawed over small sample sizes so if my eyes tell me differently, I'm not convinced that my eye test is wrong.

Over the course of a full season, I figure the opposition goaltending we face more or less evens out for all teams so goals is an accurate measure. I would guess that expected goals does pretty well too so that those two stats would order teams close to the same way. I've never checked but I suspect that's the case anyway but if not, I'd think that goals are more reliable measure.

I don't see it happening.

We got shut out for 2.9 games out of 5 and HF Faithful still want to pin the series loss on the .930 goalie

I thought it was more like 2.95 but yeah, goddamn that Andersen!

 

[egd27]

There are simple ones and more complex ones.

Simple ones like on naturalstattrick.com simply track shot location, and high danger and scoring chances are defined simply by what box on the ice those shots are from.

Other more complex scoring chance measures factor in whether shots come off rushes, rebounds, cross ice passes, and the quality of the shooter.


Of course no matter which measures you prefer, they all say the same thing about the leafs.

Who decides whether a shot is a HDC or not?

 

[zeke]

Who decides whether a shot is a HDC or not?

In the ones quoted here, from naturalstatrrick, those are simply the shot distance/location recorded by the official nhl scorer.

If they are recorded in the high danger area, they are hdc.

 

[egd27]

I thought it was more like 2.95 but yeah, goddamn that Andersen!

I rounded to 55 minutes of Game 4

 

[Gary Nylund]

I'm surprised by that number. How are 'higher danger chances' calculated/counted? It must be derived by subjective observation, no?

Unless I'm missing something then yes, it's measured by humans with all their biases and imperfections and "high danger" is subjective to begin with.

Surprised doesn't begin to cover it. I just call BS on that number, I don't trust it at all.

 

[Jimmy Firecracker]

Many are an interpretation of what occurred rather than a record of what occurred.

No different than the hits stat which many on here take as gospel.

 

[Gary Nylund]

In the ones quoted here, from naturalstatrrick, those are simply the shot distance/location recorded by the official nhl scorer.

If they are recorded in the high danger area, they are hdc.

Haha, primitive as hell. No wonder the stat is so unreliable.

 

[egd27]

In the ones quoted here, from naturalstatrrick, those are simply the shot distance/location recorded by the official nhl scorer.

So the entity deciding whether a shot is a HDC, never actually sees the shot? And I assume the official NHL scorer also varies from location to location?

 

[zeke]

No different than the hits stat which many on here take as gospel.

Actually much different.

Hits takeaways giveaways are entirely subjective and have no clear definition.

Stuff like shot distance and shot locations is objective data, though there can be questions as to its accuracy.

 

[Bomber0104]

So the entity deciding whether a shot is a HDC, never actually sees the shot? And I assume the official NHL scorer also varies from location to location?

Sounds foolproof to me

 

[Jimmy Firecracker]

Actually much different.

Hits takeaways giveaways are entirely subjective and have no clear definition.

Stuff like shot distance and shot locations is objective data, though there can be questions as to its accuracy.

True. A shot from the slot is always a shot from the slot. A hit can be anything from a massive layout to a light bump.

 

[zeke]

So the entity deciding whether a shot is a HDC, never actually sees the shot? And I assume the official NHL scorer also varies from location to location?

Yes, the ones quoted here are not anyone deciding whether a shot is high danger or not. Simply a matter of where that shot was taken from.

And yes, official scorers vary and make mistakes.

 

[zeke]

Haha, primitive as hell. No wonder the stat is so unreliable.

Who says it's unreliable?

 

[Martin Skoula]

Who says it's unreliable?

My miles per gallon changes depending on if I'm going downhill or uphill, miles per gallon is therefore a made up unreliable stat.

 

[Suntouchable13]

I think that most of us wanted our stars to play more, but Keefe is going overboard. Matthews and Marner both averaging 22 minutes of ice time a night. I am concerned if that will continue for the remainder of the season as there are no more breaks from now till the end of the season (unless we'll run into a covid situation somewhere) that will run them straight into the ground by playoff time.

 

[Mr Hockey]

So the entity deciding whether a shot is a HDC, never actually sees the shot? And I assume the official NHL scorer also varies from location to location?

They have software programs that decide all these stats, but teams have there own stat counters too and you get individuals that track games sitting at home watching on there computers or tv

 

[ShaneFalco]

Got the bounces? Explain.

He had some luck

 

[Dreakmur]

Who says it's unreliable?

Do the results actually match up with the predictions? If not, it’s unreliable.

Simple shot mapping is likely to correlate with scoring, so I would assume it’s somewhat accurate.

 

[egd27]

My miles per gallon changes depending on if I'm going downhill or uphill, miles per gallon is therefore a made up unreliable stat.

There is no interpretation on what a mile or a gallon is.

If 31 (or more) different people were deciding how far a mile was, or how big a gallon was it would be much less reliable.

 

[Dreakmur]

There is no interpretation on what a mile or a gallon is.

If 31 (or more) different people were deciding how far a mile was, or how big a gallon was it would be much less reliable.

The miles per gallon argument actually supports the scepticism for expected goals.

Under different conditions, a car’s fuel economy changes. Under different conditions, a shot’s probability of scoring changes too.

 

[Mr Hockey]

Do the results actually match up with the predictions? If not, it’s unreliable.

Simple shot mapping is likely to correlate with scoring, so I would assume it’s somewhat accurate.

look at it this way, the advanced stats probably say the Leafs will win every game until seasons end ... where/what are the flaws?

 

[Leafsfan74]

It's almost entirely derived from shot location, no account for the actual player taking the shot.

An Ovechkin bomb from the point, ceteris paribus, is valued the same as one from Luke Schenn.

The use of the word "danger" for the statistic is borderline laughable.

A more apt name for the stat should be "Closeness of Shot from the Net plus a few other trivial things".

They should be shooting more. Spezza showed why shooting is rewarded in this game with that bullet that was deflected in.

Often, too many passes just allows opposition D and goalie time to react so a " high risk chance", is less lethal.

Take a shot, get a rebound in tight, drive the net, make life difficult for the opposition goalie. I still find them lacking in these aspects. At best, they are widely inconsistent.

 

[Gallagbi]

Not sure what your point is. If it's that Campbell played well last night and with Andersen perhaps he plays worse and we lose then sure, that could have happened.

If it's that the only issue with our play during the slump is goaltending, then no.

Point is to highlight how our goaltending has been the biggest factor in the recent slump.

I don't ignore stats, they can be helpful so why not use all the tools at our disposal. On the other hand, when people post Xgoals stats for one game (or even a series) as proof of something, for me personally, I believe the stat to be flawed over small sample sizes so if my eyes tell me differently, I'm not convinced that my eye test is wrong.

This just seems like a double standard to me. Nobody posts their full picture analysis with complete backup off the start.

When using the eye test, it's rarely supported by anything more than speculation and usually surface interpretation (those are tough saves, that's a good shot).

I agree that you should use and present all factors to gain a true understanding, but can you think of a single poster who leads this way? Hell, I watched you go after the one who posts the most backup for stats yesterday

Over the course of a full season, I figure the opposition goaltending we face more or less evens out for all teams so goals is an accurate measure. I would guess that expected goals does pretty well too so that those two stats would order teams close to the same way. I've never checked but I suspect that's the case anyway but if not, I'd think that goals are more reliable measure.

Yes and no, but in general the larger the sample the better.

I agree with your earlier point on this that you'd need to look at all factors for GF to be useful. How much was 3on3, ENG, vs. backups etc .

 

[zeke]

Do the results actually match up with the predictions? If not, it’s unreliable.

Simple shot mapping is likely to correlate with scoring, so I would assume it’s somewhat accurate.

It should never match up exactly - goaltending and shooting efficiency and luck are things.

 

[Dreakmur]

It should never match up exactly - goaltending and shooting efficiency and luck are things.

How closely does it match? Should be like 80% or better to be considered reliable, right?