Brady Tkachuk is on pace for one of the greatest season in analytics history

[JetsFan815]

The guy is averaging 1.84 xGoals/60... that's insane, no one this season is even close. Matthews is a distant 2nd at 1.37/60. Tkachuk is on pace to break Ovechkin's record of the best stats in this metric at 1.81 in the 08-09 season.

Esp impressive that he's doing it on such a bad team where just getting the puck to the areas that give you good expected goals is a herculean task in itself, let alone getting those shots off.

Click to read more...

 

[7even]

Look I'm not saying Brady Tkachuk is better than McDavid. But I'm not saying he's not either.

 

[GHJimmy]

He'll eventually slow down and have slump, like any other player in the league (most)

 

[TheKrebsCycle]

Hes a shooting and hitting machine that has a nose for net. What a stud .

 

[Player big P]

Shame about the face, though

 

[Black Kevin]

I'm ready to admit it. This kid is the greatest athlete of all time.

 

[bert]

Sens fan here, just shows how meaningless some of these advanced stats are. Solid player but he isn't dominating games by any stretch of the imagination.

 

[daver]

Was skeptical of the claim until I read "in analytics history"

 

[Henkka]

The guy is averaging 1.84 xGoals/60... that's insane, no one this season is even close. Matthews is a distant 2nd at 1.37/60. Tkachuk is on pace to break Ovechkin's record of the best stats in this metric at 1.81 in the 08-09 season.

Esp impressive that he's doing it on such a bad team where just getting the puck to the areas that give you good expected goals is a herculean task in itself, let alone getting those shots off.

Tkachuk is ~4 goals goals behind (has scored 6 total goals) of that 10.18 of total expected.

Matthews has scored ~10 goals more (18 total) than his expected 8.09.

Just statistical variation or is there something flawed how these are count?

 

[Uncle Scrooge]

Im at the point with advanced stats that i just look at them and this is my brain:

Confused at first, then interested for a second... but then i realize:

 

[jbeck5]

Shame about the face, though

You wish you had his face lol

 

[Guttersniped]

Tkachuk is ~4 goals goals behind (has scored 6 total goals) of that 10.18 of total expected.

Matthews has scored ~10 goals more (18 total) than his expected 8.09.

Just statistical variation or is there something flawed how these are count?

Take a look at each players’ shooting percentages and on ice shooting percentages. That’s the variation and it’s man made.

 

[Player big P]

You wish you had his face lol

I've got enough problems

 

[bleedgreen]

Title of the thread is analytics in a nutshell to me.


And I really like the kid.

 

[Karl Eriksson]

Might be the best “Brady” in pro sports. No one else comes to mind.

 

[613Leafer]

Tkachuk is ~4 goals goals behind (has scored 6 total goals) of that 10.18 of total expected.

Matthews has scored ~10 goals more (18 total) than his expected 8.09.

Just statistical variation or is there something flawed how these are count?

Expected goals doesn't take into account the quality of the shot/shooter, whether there was traffic in front of the net, whether it was a one-timer that the goalie had to slide over for, etc.

So it's a better metric than strictly shots, but still flawed.

 

[Talonted]

Thank goodness for advanced analytics or else we wouldn’t know how good he is!

 

[iCanada]

Tkachuk is ~4 goals goals behind (has scored 6 total goals) of that 10.18 of total expected.

Matthews has scored ~10 goals more (18 total) than his expected 8.09.

Just statistical variation or is there something flawed how these are count?

They don't take into account player skill. It's just a cumulative number of shots you fire on net multiplied by a percentage chance to score depending on the location of the shot based on League averages.

You'll always see some players over perform and underperform these numbers. And it's the players that over perform them that are better. The reality is, if you let me and Ovechkin shoot the puck from his office 10 times, he likely scores more than I hit the net. But our expected goals outcome would be the same.

Like most of hockey's analytics, there is an underlying assumption that all players are equal that is patently false. I'd rather have a guy who generates one chance a game and scores on it every time than a guy who generates 15 but doesn't score. But analytics would tell you the second guy "is better" and "effects the game more."

 

[Peasy]

Tkachuk takes a shit load of shots.

Tkachuk also takes a shit load of shots near the net (banging away).

Of course his analytics will be good. Doesn't mean he's actually as good as they suggest though.

 

[Boom Boom Apathy]

Title of the thread is analytics in a nutshell to me.


And I really like the kid.

Yeah. The problem is how people use and interpret the data. I've spent a lot of my prior career doing failure life modeling of electronic components. I've seen so many cases where people can draw conclusions from data or use pieces of the data to see what they want to see, vs. recognizing what the data is saying, or understanding that the models have limitations and error bars.

In hockey, there is no single 'advanced' stat that can accurately reflect how a player is doing, particularly over a short time frame. In fact, there aren't even a collection of stats that can be the "be all, end all". On the flip side, the "eye test" also has limitations because we all have built in biases which affect how we view and interpret things, no matter how much we think we are unbiased. We also many times only see high lights that the NHL/TV networks choose to show us to make our assessments. Analytical data is useful, but it has to be used as a whole (not 1 stat) in conjunction with the "eye test" to really get value.

IIRC, Francis commented on this during his time at GM. I'm paraphrasing, but he said something to the effect of: "We use analytics to both confirm what we see, and point out things that maybe we don't see"

 

[Bomber0104]

The expected goals model heavily rewards shots close to the net.

Brady lives in front of the net looking for deflections and garbage goals (pretty much as close as it gets to the net) so it's not hard to figure out why the model grossly exaggerates how many goals he's "expected" to score.

I'm optimistic one day these stats might actually mean something.

 

[Kranix]

See this is why I would take him over McDavid right now.

 

[nobody]

They don't take into account player skill. It's just a cumulative number of shots you fire on net multiplied by a percentage chance to score depending on the location of the shot based on League averages.

You'll always see some players over perform and underperform these numbers. And it's the players that over perform them that are better. The reality is, if you let me and Ovechkin shoot the puck from his office 10 times, he likely scores more than I hit the net. But our expected goals outcome would be the same.

Like most of hockey's analytics, there is an underlying assumption that all players are equal that is patently false. I'd rather have a guy who generates one chance a game and scores on it every time than a guy who generates 15 but doesn't score. But analytics would tell you the second guy "is better" and "effects the game more."

So you're saying that you would rather take Connor McDavid over Zach Hyman? You must be a crazy person.

 

[jbeck5]

I've got enough problems

Add one lol.

 

[cudi]

Im not anti analytic at all, but the term 'expected goals' always makes me laugh.. Reminds me of the 'hypothetical dollars' from South Park.