Does Dube make the team?

[Nelson]

Originally Posted by Enoch Root

Absolutely not, and I hope I wasn't sounding like I think you were.

I just don't think that possession stats - in a 3-game, pre-season sample - can be made too much of. Highly dependent on who you're playing with and against.

The 40% shooting percentage is obviously high, and suggests things have gone well over the short number of games. But that's my point: the possession stats have to be considered with the same inapplicability, for lack of a better word. Stats simply aren't very useful in very small sample sizes.

They have both played well, and they are going to be judged on how they play as the pre-season ramps up towards regular-season intensity and quality. Regardless of any of the stats.

You’re absolutely right about the sample size warning. But what method of evaluation isn’t invalidated by the small sample size? Are we to consider ourselves ignorant about which players are performing better than others because of the small sample size? I think you can watch the players and then take into account all the stats you can find and come up with a reasonable evaluation despite the small sample size. Of course, this comes with the caveat that the evaluation is still very weak and will become stronger with games played.

[Enoch Root]

Actually, I would argue that you simply evaluate the players by watching them, and pretty much ignore the stats, because with small sample sizes they are useless.

The whole point about stats is that you need a large enough sample size to have extraneous variables average out. But in small sample sizes, those extraneous variables don't average out, they tend to dominate the numbers. Making the numbers useless.

That is a fundamental aspect of employing stats in this manner.

[GranteedEV]

Originally Posted by Enoch Root

Actually, I would argue that you simply evaluate the players by watching them

Watching is far from infallible in small sample sizes.

Who isn't guilty of confirmation bias? A player could have an actually excellent game and all I end up hearing is a fixation on minor weak areas or a lack of puck luck beig proof positive of how uselss they are. And vice versa, a player can have an awful game, score one timely goal where everyone else did 90% of the work and you'll probably hear how strong a game they had.

I'm not just talking about fans, either. Last year Tanner Glass made our team. Management is guilty of fixating on one area and overlooking the obvious.

The stats might have a small sample size but at least they give you something to dig into and examine further. You're correct that there are reasons for stats that are outside the control of a player but we can still try and see if there are reasons within the control of the player, and I hope the coaches are using that information to figure out how, especially in preseason, to set the players up for useful evaluation.

[868904]

Originally Posted by Enoch Root

Actually, I would argue that you simply evaluate the players by watching them, and pretty much ignore the stats, because with small sample sizes they are useless.

The whole point about stats is that you need a large enough sample size to have extraneous variables average out. But in small sample sizes, those extraneous variables don't average out, they tend to dominate the numbers. Making the numbers useless.

That is a fundamental aspect of employing stats in this manner.

But the stats that Bingo alluded to are supported by the eye test too. Dube has been great at generating offense and is producing too, but he's also had some big defensive miscues, whereas we haven't really noticed any defensive miscues from Mangiapane and Foo.

I don't think anyone is saying that Dube isn't our best forward prospect or best prospect period. He is clearly projecting better than Mangiapane and Foo but it also shouldn't be surprising that both Mangiapane and Foo are managing the puck better defensively considering the pro experience they both have.

There really isn't any harm in sending all 3 of those guys down to Stockton for seasoning. There will be injuries all year long and you need to maintain depth. Flames need to be 6 lines and pairings deep throughout the NHL and AHL rosters combined to have success. If you can marinate Dube, Foo and Mangiapane in the AHL for a bit more and keep Hathaway and Lazar, then you do what you can.

[Enoch Root]

I agree that Dube has had a couple terrible defensive miscues. But I actually also think he has made more good defensive plays than Mangiapane, and has shown to be a smarter all around player.

Having said that, I agree that all 3 of those guys, as well as the organization, are best served with them in Stockton. What would be best for the team, IMO, is one of Rychel, Lazar or Hathaway show that they deserve to make the top 12, so that these 3 can play with the Heat.

But so far (it's still very early), none of those guys have done that.

[Bingo]

Originally Posted by Enoch Root

Actually, I would argue that you simply evaluate the players by watching them, and pretty much ignore the stats, because with small sample sizes they are useless.

The whole point about stats is that you need a large enough sample size to have extraneous variables average out. But in small sample sizes, those extraneous variables don't average out, they tend to dominate the numbers. Making the numbers useless.

That is a fundamental aspect of employing stats in this manner.

The eye test is what sent me to look up the stats in the first place though. I get a bad rap from the anti-stats crowd, but I'm a hockey guy through and through and I watch the games closely.

Mikael Backlund is the first player for me that had the stats come along and back up the eye test for me. He was doing all the little things right that you almost never see on the game sheet at the end of the night.

Mangiapane and Foo just don't get in much trouble, and that stands out to me as the game has gone away from offensive specialists, or at least it's trying too.

Their games are more mature.

And the stats in this case are just tallying events on the ice, done by an independent set of eyes that have no skin in the game. If you tend to never give up chances against when you're on the ice, the tallying says so, I don't think the coaching eye is far behind in seeing the same thing.

[Enoch Root]

Originally Posted by GranteedEV

Watching is far from infallible in small sample sizes.

Who isn't guilty of confirmation bias? A player could have an actually excellent game and all I end up hearing is a fixation on minor weak areas or a lack of puck luck beig proof positive of how uselss they are. And vice versa, a player can have an awful game, score one timely goal where everyone else did 90% of the work and you'll probably hear how strong a game they had.

I'm not just talking about fans, either. Last year Tanner Glass made our team. Management is guilty of fixating on one area and overlooking the obvious.

The stats might have a small sample size but at least they give you something to dig into and examine further. You're correct that there are reasons for stats that are outside the control of a player but we can still try and see if there are reasons within the control of the player, and I hope the coaches are using that information to figure out how, especially in preseason, to set the players up for useful evaluation.

Of course watching is far from infallible.

But if the coaches can't sit down and make good evaluations of the players, from watching the film, then they would never have made it to become NHL coaches in the first place.

I want coaches that evaluate players by watching the film. Not guys looking at spreadsheets. Especially spreadsheets from pre-season games where some of the players on the ice can't even play in the A.

[Enoch Root]

Originally Posted by Bingo

The eye test is what sent me to look up the stats in the first place though. I get a bad rap from the anti-stats crowd, but I'm a hockey guy through and through and I watch the games closely.

Mikael Backlund is the first player for me that had the stats come along and back up the eye test for me. He was doing all the little things right that you almost never see on the game sheet at the end of the night.

Mangiapane and Foo just don't get in much trouble, and that stands out to me as the game has gone away from offensive specialists, or at least it's trying too.

Their games are more mature.

And the stats in this case are just tallying events on the ice, done by an independent set of eyes that have no skin in the game. If you tend to never give up chances against when you're on the ice, the tallying says so, I don't think the coaching eye is far behind in seeing the same thing.

To be clear, Bingo. I am not trying to give you a bad rap. I would argue that you present and use the stats as well as anyone. Which is precisely why I engage you on them - you can present the other side well.

It's the stats that I have a problem with. Too much noise in possession stats. And too many people that extract too much confidence and conviction from the numbers.

(and this comes from someone who made a career of analyzing and applying stats)

[pepper24]

If Foo and Mangiapane are more mature and right up there with Dube I like that problem. 3 players that can step in whether opening roster or call ups.

Makes Hathaway/Lazar 13th forward or expendable in short term. Gives Flames leverage with Bennett being last year of show me deal and pushed back in depth charts.

[Bingo]

Originally Posted by Enoch Root

To be clear, Bingo. I am not trying to give you a bad rap. I would argue that you present and use the stats as well as anyone. Which is precisely why I engage you on them - you can present the other side well.

It's the stats that I have a problem with. Too much noise in possession stats. And too many people that extract too much confidence and conviction from the numbers.

(and this comes from someone who made a career of analyzing and applying stats)

I certainly agree with that.

Straight corsi especially has the danger of running up events in trivial ways to get a positive CF% without really accomplishing anything.

I'm on the don't discount, but don't go only on stats bandwagon, and I think most NHL coaches are the same. They have a "feel" then they check the stats to see if they match. If they don't you watch the tape again and see why.

[DeluxeMoustache]

Originally Posted by Enoch Root

To be clear, Bingo. I am not trying to give you a bad rap. I would argue that you present and use the stats as well as anyone. Which is precisely why I engage you on them - you can present the other side well.

It's the stats that I have a problem with. Too much noise in possession stats. And too many people that extract too much confidence and conviction from the numbers.

(and this comes from someone who made a career of analyzing and applying stats)

I definitely agree with this.

Making a career analyzing and applying stats, that naturally would result in understanding statistical significance, understanding the limitations, etc. (Side note: single game sv% when people note it makes me roll my eyes)

One challenge with using a lot of counting stats for individuals is that you can’t really practically validate the result against another set of data, and the result can’t become predictive. And when looking at this, you isolate the impact of a player from the other 11 guys on the ice, but you can’t add back in the other 11 in any model and get a meaningful predictive result.

And for teams, trying to correlate possession metrics to team success, generally it makes sense but you only need to look at the last 4 years here to see two results that blatantly defy that correlation.

[edslunch]

Stats aside, I think promoting players directly from junior to the NHL is a bad idea unless they are clearly too good for junior and the AHL is not an option (e.g. typical high first round pick from CHL). Otherwise you’re rushing their development at the risk of stunting it, just to get them some 4th line minutes.

As exciting as Dube is as a prospect he’s far from a finished product. The Flames don’t need him in the NHL this year, so send him down and him learn and hopefully dominate in the AHL for a year.

[CroFlames]

Originally Posted by Enoch Root

To be clear, Bingo. I am not trying to give you a bad rap. I would argue that you present and use the stats as well as anyone. Which is precisely why I engage you on them - you can present the other side well.

It's the stats that I have a problem with. Too much noise in possession stats. And too many people that extract too much confidence and conviction from the numbers.

(and this comes from someone who made a career of analyzing and applying stats)

The stats Bingo is using is that when Dube is on the ice, more chances are given up compared to Foo or Mangiapane.

Watching tape would confirm that pretty easily, and the explanation makes logical sense too: Dube has very minimal professional experience.

No doubt in my mind he will learn to take care of housekeeping in the future, and that future may be as soon as this season. Time will tell, but Bingo certainly isn't making any conclusions off advanced stats like corsi here.

[belsarius]

Young guy with lots of offensive upside and drive to make the team but needs work on the defensive side of the puck?

Paging Mr. Backlund, your talents are required.

[Geeoff]

I just realized that if Dube makes the team, that means no Lazar OR Hathaway and every player on the team actually has a chance to score if they take a shot. O_O

[442scotty]

Dube is too good for the ahl. He should stay up and learn to refine his game on the fly which won’t take long.

[Bingo]

Originally Posted by DeluxeMoustache

I definitely agree with this.

Making a career analyzing and applying stats, that naturally would result in understanding statistical significance, understanding the limitations, etc. (Side note: single game sv% when people note it makes me roll my eyes)

One challenge with using a lot of counting stats for individuals is that you can’t really practically validate the result against another set of data, and the result can’t become predictive. And when looking at this, you isolate the impact of a player from the other 11 guys on the ice, but you can’t add back in the other 11 in any model and get a meaningful predictive result.

And for teams, trying to correlate possession metrics to team success, generally it makes sense but you only need to look at the last 4 years here to see two results that blatantly defy that correlation.

I think you're over thinking things a bit. This isn't a model to predict a meteor strike on Earth in the next 30 years.

It's a guy counting how many times a guy tries a shot in both directions with what players on the ice.

Does it 100% correlate to anything in a season? Usually to 85%. In 4 preseason games of course not. But remember the coaching staff is trying to make all their decisions on 4-5 preseason games, the sample size is small every year.

If player A plays 45 minutes and gives up only a handful of chances against versus player B that seems to be playing riverboat hockey ... I'd would imagine they'd take that into consideration.

Heading into tonight's game ...

Magiapane is 28-7 in scoring chances, and with 43 minutes of ice time that's getting pretty interesting. (12-1 in high danger).

Dube is 23-21 and 12-7 in 46 minutes

[Royle9]

How does one calculate chances against though?
Would 28-7 for Andrew be solely on him being great and reducing the chances or is it on the entire line that's on the ice as well as offensive vs defensive zone ice time etc? Also, he has 28 chances and a large part of those were playing on a line with Dube, including his only goal SH.

My eye test shows me that Dube's looked more dangerous than most players on the ice and his skating/puck work is producing more than most as well.


I'd like to know how this chances for/against is actually calculated because it seems odd to me honestly.
I'm no stat's expert and I'll be the first to admit it but in talking with my circle of hockey buddies here at work everyone seems to agree that Dube has looked better this pre-season including non flames fans and advance stat junkies.

[Bingo]

Originally Posted by Royle9

How does one calculate chances against though?
Would 28-7 for Andrew be solely on him being great and reducing the chances or is it on the entire line that's on the ice as well as offensive vs defensive zone ice time etc?

My eye test shows me that Dube's looked more dangerous than most players on the ice and his skating/puck work is producing more than most as well.


I'd like to know how this chances for/against is actually calculated because it seems odd to me honestly.

Just being on the ice.

Mangiapane could be the least interested defensive player in camp but have played with elite linemates that have done all the work and he's the benefactor of being along for the ride.

Don't think that's the case but it's possible.

Similarly Dube could be doing everything right but playing with linemates that keep screwing up. Possible, but don't see that either.

Dube has opened my eyes too guys, his speed and vision have me thinking top six which isn't something I was pondering prior to this year's camp. But just like you're seeing in Edmonton right now with Yamamoto and Rattie, the first four preseason games just don't matter to me, so people handing out stats based on a hat trick against the Condors gets under my skin.

I hope one of Mangiapane or Dube rise to the occasion this week and make the team. I'm just not ready to hand out spots based on games to this point.

[JohnnyTitan]

I want Dube up on the big club as I think he makes our bottom-six super-dynamic with Bennett / Janko / Ryan / Czarnick / Frolik. Wow.

Just devil's advocate here...but how deep and cost-controlled does our team look like with essentially all our forwards and d-men locked-up next year, Lazar / Kulak as our #13 / #7 and this in the AHL:
Mangipane - Dube - Foo
Valimaki - Kylington
Parsons - Gillies

Well done Mr. Treliving!

And this coming from an optimist expecting Ryan to deliver exactly what he's supposed to and Bennett / Janko / Czarnick / Dube to all "Pop".

Gonna be a great few years!