Canucks Analytics

[sting101]

BEN HUTTON (VAN) - 5v5

WITH ERIK GUDBRANSON
TOI	SAT For	SAT Against	SAT Diff	SAT %	SOG For	SOG Against	SOG Diff	SOG %	Goals For	Goals Against	+/-	GF %	OZ FO	NZ FO	DZ FO	OZ%
467:04	453	608	-155	42.7	252	335	-83	42.93	17	36	-19	32.08	188	176	199	48.58

[TBODY] [/TBODY]
[TBODY] [/TBODY]

WITHOUT ERIK GUDBRANSON
TOI	SAT For	SAT Against	SAT Diff	SAT %	SOG For	SOG Against	SOG Diff	SOG %	Goals For	Goals Against	+/-	GF %	OZ FO	NZ FO	DZ FO	OZ%
326:27	370	358	12	50.82	141	153	-12	47.96	14	8	6	63.64	74	78	66	52.86

[TBODY] [/TBODY]
[TBODY] [/TBODY]

thought this one is pretty funny

 

[Melvin]

I've updated the WOWY tool to include last season. Now you can do Hutton wow Gudbranson over both years:

I will work on going back and probably adding one season per day until I have everything.

EDIT: Some of the numbers in the above screenshot are incorrect. I have since fixed a couple bugs with the 5v5 data. TOI without Gudbranson is much higher.

 

[Bojack Horvatman]

BEN HUTTON (VAN) - 5v5

WITH ERIK GUDBRANSON
TOI	SAT For	SAT Against	SAT Diff	SAT %	SOG For	SOG Against	SOG Diff	SOG %	Goals For	Goals Against	+/-	GF %	OZ FO	NZ FO	DZ FO	OZ%
467:04	453	608	-155	42.7	252	335	-83	42.93	17	36	-19	32.08	188	176	199	48.58

[TBODY] [/TBODY]
[TBODY] [/TBODY]

WITHOUT ERIK GUDBRANSON
TOI	SAT For	SAT Against	SAT Diff	SAT %	SOG For	SOG Against	SOG Diff	SOG %	Goals For	Goals Against	+/-	GF %	OZ FO	NZ FO	DZ FO	OZ%
326:27	370	358	12	50.82	141	153	-12	47.96	14	8	6	63.64	74	78	66	52.86

[TBODY] [/TBODY]
[TBODY] [/TBODY]

thought this one is pretty funny

If Benning was smart he would buy him out this summer as he would be a detriment to the development of Hutton, Hughes and Juolevi. He is a gift for team tank though so, I hope he stays a Canuck.

 

[tyhee]

Thanks Melvin. Great tool.

After only about 10 uses, the only issue I've had is that on one occasion it timed out on me after 30 seconds.

 

[Melvin]

Thanks!

Yeah the performance was temporarily very bad after I added an entire season's worth of data, but I think I've sorted that out.

 

[Melvin]

I've added another page for breaking down the shots attempted by all players. Some examples of things you can look at:

Troy Stecher leads the team with 43% of his shot attempts being blocked.
Meanwhile, Roussel has missed the net with 31% of his attempts.
Edler has taken the most Slapshots; Horvat has attempted a team-leading 7 wrap-arounds.
Schaller's attempts have come from an average of 25 feet away, which is the nearest on the team; Tanev's are the longest distance attempts with an average of 68 ft.

NHL Tools

Still just the tip of the iceberg really. There is still so much potential in terms of things to look at and add. Really excited to finally have the time to put stuff like this together.

 

[Bleach Clean]

I had been missing the "With or Without You" tool that stats.hockeyanalysis used to provide, so I finally took the time to build one myself.

Got a first prototype up and running now, so if anyone wants to try it out I'd appreciate it. URL is NHL Tools - With or Without You .

Simply select the team in the top-left, load the players, then select the two players to compare and it will show you their on-ice stats when together and when apart.

It is very bare-bones as I just got it up and running today, but the data should be accurate so if you want to try it out I could use some testing/feedback.

IF you can think of any stuff to add, whether it's other stats or it's other tools please let me know as well.

Where are you getting the data for this? CSV files?

Hutton vs. Pouliot numbers amuse me.

 

[HankNDank]

Bo Horvat with Gudbranson - 36.4% GF%
Bo Horvat without Gudbranson - 55.6 GF%

Elias Pettersson with Gudbranson - 47.1 GF%
Elias Pettersson without Gudbranson - 67.5 GF%

Gudbranson can even turn the almighty Pettersson into an average hockey player.

 

[Melvin]

Where are you getting the data for this? CSV files?

Hutton vs. Pouliot numbers amuse me.

I get it straight from the NHL.com game reports, e.g. Play By Play

 

[4Twenty]

Why wouldn't you just use naturalstattrick.com, the works already done.

 

[Melvin]

Why wouldn't you just use naturalstattrick.com, the works already done.

That's a good attitude. Your handle checks out.

I'm not going to steal their work.

 

[4Twenty]

That's a good attitude. Your handle checks out.

I'm not going to steal their work.

I don't get it?

You can look at with and without stats right on their page.

 

[Melvin]

I don't get it?

You can look at with and without stats right on their page.

First of all, I didn't know that. Never could find WOWY stats on their page.

Second of all, I want the raw data for my own purposes. That I quickly threw a website around it so that other people could look at it is just a by product.

Anyway, it's always good to have secondary sources for cross-validation. If you don't like it don't use it. [MOD]

 

[4Twenty]

First of all, I didn't know that. Never could find WOWY stats on their page.

Second of all, I want the raw data for my own purposes. That I quickly threw a website around it so that other people could look at it is just a by product.

Anyway, it's always good to have secondary sources for cross-validation. If you don't like it don't use it. [MOD].

You must be having a bad day to suggest I was being a dick about anything. Chill out my man.

You do good work, I appreciate it, I just thought suggesting a public site would be welcome to you. I'm not diminishing what you're doing.

Nor do I see any way which anyone could deem my post as being a dick.....you're being overly sensitive.

I literally clicked on the thread, first post I saw was you or someone lamenting a lack of with/without stats. I thought pointing out naturalstattrick has the info you missed was a positive, [MOD]

 

[Melvin]

Sorry. You're right, I am in a bad mood. It's my birthday and I'm super hungover and irritated.

 

[4Twenty]

Happy Birthday.

 

[Bleach Clean]

I get it straight from the NHL.com game reports, e.g. Play By Play

How are you grabbing and parsing this information? Do you use a script + SQL + Dbase?


------------------------------------------





Also:

 

[Melvin]

Yeah I wrote some parsers in Python and I have a database in Azure

 

[Melvin]

I thought this might be interesting.

Take each team's goal +/- at 5v5 by the total amount of minutes played per player. Then you can determine the "expected" +/- of the player given the team for whom he's played and how many minutes. Then you can look at a player's actual +/- to give you a more context-neutral way of looking at goals on the ice for. Not perfect for a whole plethora of obvious reasons but does make a bit more sense than raw +/-.

So for example Troy Stecher's +14 at 5v5(*) doesn't look like anything special, but given his minutes and team, and his expected +/- is actually -2, so his +16 "adjusted +/-" actually ranks him 10th in the NHL among all players, and 3rd among defenders. I dunno, I think it's interesting.

		Minutes	Expected +/-	Actual +/-	Adjust +/-
PIT	SIDNEY CROSBY	800	7	31	24
OTT	MARK STONE	818	-9	13	22
OTT	DYLAN DEMELO	764	-9	12	21
DAL	TYLER SEGUIN	827	-1	18	19
BUF	JEFF SKINNER	776	-1	17	18
CAR	SEBASTIAN AHO	754	-1	17	18
COL	MIKKO RANTANEN	870	-2	14	16
MIN	ERIC STAAL	720	-2	14	16
S.J	BRENDEN DILLON	805	1	17	16
VAN	TROY STECHER	739	-2	14	16

[TBODY] [/TBODY]

Just thought I'd share.

(*) Can we talk about the fact that Stecher has spent 50 minutes killing penalties, allowed just one goal and been on the ice for 2 shorties? Like he's a +1 in short-handed situations! Anyway.

A confession: I am not completely sold on using on-ice shot data exclusively for evaluations. I think there is still value in looking at goals. As a data professional who has been doing analytics in various industries for the past 10 years, I find it very frustrating that "analytics" in hockey has become synonymous with on-ice shot attempts (stupidly called "corsi") which is just a metric and not all-encompassing of all analytics (which just means to study data.) I think that a lot of people have made the mistake of assuming certain correlations without sufficiently evidencing it and it kind of drives me nuts sometimes. Rant over.

Oh, and because I'm sure you weree wondering:

CBJ	BRANDON DUBINSKY	390	0	-15	-15
VAN	ERIK GUDBRANSON	729	-2	-21	-19
WSH	MATT NISKANEN	854	5	-15	-20
WSH	DMITRY ORLOV	892	5	-16	-21
PIT	JACK JOHNSON	800	7	-15	-22

[TBODY] [/TBODY]

 

[Phenomenon13]

Bo Horvat with Gudbranson - 36.4% GF%
Bo Horvat without Gudbranson - 55.6 GF%

Elias Pettersson with Gudbranson - 47.1 GF%
Elias Pettersson without Gudbranson - 67.5 GF%

Gudbranson can even turn the almighty Pettersson into an average hockey player.

Sounds perfect for our tanking

 

[Melvin]

Been playing lately with shot distance data. If you look at how often goals are scored on attempts based on their distance from the net it looks like this:

Nothing too surprising of course; the further from the net you are, the less likely your shot is going to be a goal.

What I like about it is that it allows us to look at from where players take their shots and grade them in terms of how often they score vs. how often all players score from that distance. This shows us who the best shooters are in terms of converting on shots from distances.

Unsurprisingly, Elias Pettersson is top-5 in the NHL in this regard, and Boeser just outside the top-10:

Player	Unblocked Shots	Avg. Distance (ft.)	Goals	Exp. Goals	+/-
ALEX OVECHKIN	168	32.58	22	11.11	10.89
JOHN TAVARES	176	23.95	24	14.45	9.55
LEON DRAISAITL	111	29.24	18	8.50	9.50
JAKE GUENTZEL	155	26.83	21	11.89	9.11
ELIAS PETTERSSON	75	29.36	14	5.46	8.54
JEFF SKINNER	154	28.06	20	11.91	8.09
VIKTOR ARVIDSSON	115	29.12	16	8.47	7.53
RYAN DZINGEL	112	27.79	16	8.68	7.32
STEVEN STAMKOS	122	32.1	15	7.74	7.26
CAM ATKINSON	183	33.45	19	11.95	7.05
BROCK BOESER	92	35.05	12	5.57	6.43

[TBODY] [/TBODY]

Tavares is expected to have more goals than OV because he tends to take shots from closer. Both OV and Boeser average 30+ feet on their shots whereas guys like Tavares and Guentzel tend to shoot from closer. EP40 on the other hand has only made 75 attempts and is expected to only have around 5 goals but instead has 14 (at 5v5.)

In terms of how all the Canucks score, it looks like this:

Player	Unblocked Shots	Avg. Distance (ft.)	Goals	Exp. Goals	+/-
ELIAS PETTERSSON	75	29.36	14	5.46	8.54
BROCK BOESER	92	35.05	12	5.57	6.43
SVEN BAERTSCHI	35	22.23	5	3.16	1.84
JOSH LEIVO	48	28.46	5	3.58	1.42
BRENDAN GAUNCE	2	33.5	1	0.11	0.89
LOUI ERIKSSON	84	27.77	7	6.32	0.68
ADAM GAUDETTE	36	33.31	3	2.44	0.56
BEN HUTTON	93	55.95	3	2.47	0.53
MARKUS GRANLUND	78	31.42	6	5.68	0.32
DARREN ARCHIBALD	9	34.89	1	0.70	0.30
ALEXANDER EDLER	80	62.71	2	1.72	0.28
BRENDAN LEIPSIC	31	38.58	2	1.77	0.23
JAKE VIRTANEN	135	32.27	9	8.94	0.06
CHRISTOPHER TANEV	52	64.77	1	1.05	-0.05
ERIK GUDBRANSON	85	57.09	2	2.28	-0.28
MICHAEL DEL ZOTTO	34	49.53	1	1.39	-0.39
BRANDON SUTTER	39	33	2	2.55	-0.55
NIKOLAY GOLDOBIN	83	30.27	5	5.73	-0.73
ANTOINE ROUSSEL	81	34.02	5	5.76	-0.76
BO HORVAT	141	27.66	10	10.98	-0.98
DERRICK POULIOT	65	50.03	1	2.22	-1.22
SAM GAGNER	19	28.95	0	1.43	-1.43
TROY STECHER	75	50.25	1	2.47	-1.47
ALEX BIEGA	45	54	0	1.56	-1.56
TYLER MOTTE	88	34.4	4	5.84	-1.84
JAY BEAGLE	24	25.58	0	1.97	-1.97
TIM SCHALLER	38	24.37	0	3.14	-3.14

[TBODY] [/TBODY]

This is all 5v5; I haven't look into any PP data yet to see if the curve is any different. That's probably next.

 

[Hockeyphysio]

Awesome Melvin really enjoying it.

 

[Peter10]

This is pretty good stuff Melvin!

 

[Jay Cee]

Thanks for the work Melvin. It's interesting to see a big picture view of things instead of cherry picking a couple of stats and saying they are the be all and end all of hockey.

 

[Tables of Stats]

I wonder if there's any way to extract data about both shots that go in as well as shots that generate additional scoring chances (ie. a shot and a rebound). I think some shooters that aren't skilled enough to beat a goalie clean must be better at generating second chances with their shot than others. From there you could generate an expected value per shot and track players by that metric, find out who actually turns shot attempts into points, and who just shoots uselessly generating 'possession stats' but not actually creating value.

It might turn out that finding what I'll dub 'shooting playmakers' is a way to further identify high-value players via stats.