RGI - Index to Measure players Grit and competiveness

[ricardodw]

After taking heat from using low PIM as an indicator of a soft player I thought how can I use stats to show the grittiness (or lack thereof) .
I thought that I could show that there are no successful teams that had as many soft players as the Flames and Oilers.

I came up with Ricardo’s Grit Index (RGI). I take the plays which require grit and physical sacrifice (hits, blocked shots and Take-aways) and subtract Give-aways to get a raw grit score.

I then do a simple grit / games played . This does not show how gritty a player is relative to his peers. Someone like Bouwmeester playing 25 minutes/game has over twice the opportunity to make gritty plays than Jackman. So I took the TOI and normalized it to regular games. A average forward should get 15 min/game of ice time (12 forwards dressed) and an average D-man should get 20 min/game of ice time. The Total TOI /15 for forwards and Total TOI/20 gives something I call regular games played.

I ended up with some surprising results.

Baertschi who I had classified as tiny skilled play scored 1.2 /game played and 1.3 for regular game played on the RGI. This was driven by his 12 Take-aways versus 5 give-aways. Statistically he is strong on the puck.

Glencross, who is one of my favorite players, scored really low in my estimation with 1.5 /game and 1.2/ reg game.

Butler has far grittier numbers than anyone would guess.. His 53 hits and 74 blocked shots made up for his poor Giveaways and left him with a 3.1 /reg game rating second only to Sarich on the Flames Defense (Cundari is up there but very small sample)

At the bottom the Flames had 4 regular forwards below 1.0: Cammalleri, Tanguay, Hudler and Cervenka……. The Flames have addressed this (bye bye Tanguay and Cervenka)…. If anyone is looking for a statistical analysis as to why Edmonton with world class talent will never make the playoffs see below

player -- grit / gm -- grit/reg game
Tim Jackman -- 2.7 -- 5.3
Carter Bancks -- 5.0 -- 5.1
Brian McGrattan -- 2.0 -- 4.2
Steve Begin -- 2.2 -- 4.2
Mark Cundari -- 4.0 -- 4.1
Cory Sarich -- 2.5 -- 3.3
Blair Jones -- 2.3 -- 3.2
Chris Butler -- 2.6 -- 3.1
Mark Giordano -- 3.4 -- 2.9
Blake Comeau -- 2.4 -- 2.9
Brett Carson -- 1.3 -- 2.4
Ben Street -- 2.2 -- 2.4
Dennis Wideman -- 2.6 -- 2.1
Roman Horak -- 1.9 -- 2.0
Mikael Backlund -- 1.9 -- 1.9
TJ Brodie -- 1.9 -- 1.8
Derek Smith -- 1.1 -- 1.8
Matt Stajan -- 2.0 -- 1.7
Akim Aliu -- 1.2 -- 1.6
Jarome Iginla -- 2.1 -- 1.6
Ben Hanowski -- 1.4 -- 1.6
Jay Bouwmeester -- 2.0 -- 1.6
Sven Baertschi -- 1.2 -- 1.3
Maxwell Reinhart -- 1.3 -- 1.3
Lee Stempniak -- 1.5 -- 1.3
Anton Babchuk -- 0.7 -- 1.2
Curtis Glencross -- 1.5 -- 1.2
Paul Byron -- 0.8 -- 1.1
Alex Tanguay -- 1.0 -- 0.7
Mike Cammalleri -- 0.8 -- 0.7
Jiri Hudler -- 0.6 -- 0.5
Roman Cervenka -- 0.3 -- 0.3

Compared to Chicago With 1 player under 1.0

player -- grit / gm -- grit/reg game
Brandon Bollig -- 1.9 -- 3.6
Brent Seabrook -- 3.9 -- 3.6
Jamal Mayers -- 1.6 -- 3.5
Daniel Carcillo -- 2.0 -- 3.4
Sheldon Brookbank -- 2.1 -- 3.4
Bryan Bickell -- 2.8 -- 3.2
Michal Rozsival -- 2.6 -- 2.8
Michal Handzus -- 2.1 -- 2.6
Michael Frolik -- 2.0 -- 2.4
Andrew Shaw -- 1.8 -- 2.4
Niklas Hjalmarsson -- 2.4 -- 2.3
Jimmy Hayes -- 2.1 -- 2.2
Johnny Oduya -- 2.0 -- 2.0
Dave Bolland -- 1.9 -- 1.7
Nick Leddy -- 1.5 -- 1.7
Viktor Stalberg -- 1.4 -- 1.5
Duncan Keith -- 1.7 -- 1.5
Marcus Kruger -- 1.3 -- 1.3
Jonathan Toews -- 1.7 -- 1.3
Brandon Saad -- 1.4 -- 1.3
Marian Hossa -- 1.5 -- 1.2
Patrick Sharp -- 1.4 -- 1.1
Patrick Kane -- 0.7 -- 0.5

Oilers are in deep deep hole that they refuse to address.
player -- grit / gm -- grit/reg game
Ben Eager -- 3.4 -- 7.0
Mark Fistric -- 5.3 -- 6.9
Ladislav Smid -- 5.3 -- 5.2
Mike Brown -- 2.8 -- 4.8
Jeff Petry -- 3.8 -- 3.5
Nick Schultz -- 2.9 -- 3.1
Lennart Petrell -- 2.3 -- 3.0
Chris VandeVelde -- 1.4 -- 2.9
Ryan Jones -- 2.4 -- 2.9
Corey Potter -- 2.2 -- 2.6
Jerred Smithson -- 2.0 -- 2.5
Teemu Hartikainen -- 1.3 -- 1.8
Eric Belanger -- 1.4 -- 1.5
Ryan Whitney -- 1.3 -- 1.4
Ryan Nugent-Hopkins -- 1.6 -- 1.3
Magnus Paajarvi -- 1.1 -- 1.2
Anton Lander -- 0.6 -- 1.1
Ryan Smyth -- 1.1 -- 1.1
Nail Yakupov -- 1.0 -- 1.1
Shawn Horcoff -- 1.0 -- 0.9
Sam Gagner -- 1.1 -- 0.9
Taylor Hall -- 1.0 -- 0.8
Jordan Eberle -- 0.9 -- 0.7
Ales Hemsky -- 0.5 -- 0.6
Justin Schultz -- 0.6 -- 0.6

A sample of other teams non-gritty players:
Sharks (4) Marleau, Thornton, Havlat (if you get paid that much you shouldn’t have to hit or block shots) and Gomez.
Canucks(4) Sedins (2), Mason Raymond and Roy … adding Roy for the playoffs!!!!

[kirant]

Interesting metric.

Assuming this is done in a spreadsheet, could you try doing an analysis of teams which have been conference finalists vs average? I think there could be some potential with this value, but we'd need more teams to determine if there's a huge impact to success. All that we can say here is that the Flames have a mean around 2 or so, the Oilers probably closer to 1.7, and the Blackhawks around the same as the Flames, but maybe a bit more evenly distributed

Also, I imagine there is some subjectivity in this stat. Unlike like shots on goal, points, or Corsi stats, the value of hits, giveaways, and takeaways are up to the recorder as to what qualifies as one.

[Itse]

Not the worst idea I've seen. However, I would have one nitpick.

Originally Posted by ricardodw

A average forward should get 15 min/game of ice time (12 forwards dressed) and an average D-man should get 20 min/game of ice time. The Total TOI /15 for forwards and Total TOI/20 gives something I call regular games played.

If you're using different metrics for defensemen and forwards, I don't think you should put them on the same list.

Also, I don't see why you don't just count it by the minute and use a multiplier if you want to make bad fractions go away.

[Vulcan]

One other nitpick, hits in one building are just incidental contact in another so comparing teams isn't going to be exact, but a good start and effort.

Just to enforce your previous argument about penalty minutes being indicative of aggressiveness, I've heard this used when evaluating players but it doesn't beat live watching. So you see some player with Lady Byng stats, so you watch to see, does he play like a wimp or is he just smart and skilled so he doesn't need to cheat.

[Cali Panthers Fan]

Originally Posted by ricardodw

Butler has far grittier numbers than anyone would guess.. His 53 hits and 74 blocked shots made up for his poor Giveaways and left him with a 3.1 /reg game rating second only to Sarich on the Flames Defense (Cundari is up there but very small sample)

player -- grit / gm -- grit/reg game
Tim Jackman -- 2.7 -- 5.3
Carter Bancks -- 5.0 -- 5.1
Brian McGrattan -- 2.0 -- 4.2
Steve Begin -- 2.2 -- 4.2
Mark Cundari -- 4.0 -- 4.1
Cory Sarich -- 2.5 -- 3.3
Blair Jones -- 2.3 -- 3.2
Chris Butler -- 2.6 -- 3.1
Mark Giordano -- 3.4 -- 2.9
Blake Comeau -- 2.4 -- 2.9
Brett Carson -- 1.3 -- 2.4
Ben Street -- 2.2 -- 2.4
Dennis Wideman -- 2.6 -- 2.1
Roman Horak -- 1.9 -- 2.0
Mikael Backlund -- 1.9 -- 1.9
TJ Brodie -- 1.9 -- 1.8
Derek Smith -- 1.1 -- 1.8
Matt Stajan -- 2.0 -- 1.7
Akim Aliu -- 1.2 -- 1.6
Jarome Iginla -- 2.1 -- 1.6
Ben Hanowski -- 1.4 -- 1.6
Jay Bouwmeester -- 2.0 -- 1.6
Sven Baertschi -- 1.2 -- 1.3
Maxwell Reinhart -- 1.3 -- 1.3
Lee Stempniak -- 1.5 -- 1.3
Anton Babchuk -- 0.7 -- 1.2
Curtis Glencross -- 1.5 -- 1.2
Paul Byron -- 0.8 -- 1.1
Alex Tanguay -- 1.0 -- 0.7
Mike Cammalleri -- 0.8 -- 0.7
Jiri Hudler -- 0.6 -- 0.5
Roman Cervenka -- 0.3 -- 0.3

Butlersoft is not at the bottom. Your system is flawed.

[ricardodw]

In SOB's crappy season last year in Colorado he had a RGI of 3.5 -- basically a wash with Sarich's 3.3

David Jones has a RGI of 1.7 which sort of means that Iginla's grit of 1.6 won't be missed. I was surprised that Iginla was not hitting that much ... I guess the union brother's figured out not to get in his face to wake him up.


TJ Galiardi has a RGI of 1.7.....

Kris Russell had a 3.2 for St.L.

St.L had 1 forward below 1.0.... MacDonald at .4 replaced by Roy.
Bouwmeester was their softest D-man in his 14 games with the Blues with a 1.2 RGI

Perron was one of St.L least gritty forwards with a 1.2 --- same as Paajarvi who was traded for him.

[ricardodw]

Originally Posted by Cali Flames Fan

Butlersoft is not at the bottom. Your system is flawed.

I had to look twice.... The man is a hitting ... shot blocking machine, If he could be a bit stronger on the puck who knows how good he would be.

I suspect that the Flames coaches and management have some sort of toughness index that Butler shows well on... I know I never got a $450k raise.

[Cali Panthers Fan]

Originally Posted by ricardodw

I had to look twice.... The man is a hitting ... shot blocking machine, If he could be a bit stronger on the puck who knows how good he would be.

I suspect that the Flames coaches and management have some sort of toughness index that Butler shows well on... I know I never got a $450k raise.

I don't doubt that he's solid on those things, but he's constantly in the wrong spot. Not a highly intelligent player.

[Justin Azevedo]

this is one of the least-needed threads in the history of the internet.

takeaways are inherently good, but the situation which leads to them is bad. this is the reason guys like datsyuk and crosby have the most giveaways in the league - it's because they always have the puck on their stick. if a guy has a lot of takeaways, it suggests that most of the time when he's on the ice he's chasing the puck, not possessing it.

hits, takeaways, giveaways and blocks are very very subjective. they fluctuate by a huge margin rink to rink as there is no official definition for what constitutes a block, hit, etc.

more pims does not make a player more gritty or better in any way. the only thing penalties accomplish is hurting the team.

although i'm sure all of these points will be missed, this horrific, illogical attempt at the evaluation of players necessitated a response.

[SebC]

When comparing teams, you wouldn't want to normalize minutes and add them all up. You'd just take total hits + blocked shots + takeaways - giveaways.

[Pierre "Monster" McGuire]

Good job Ricardo. Nice post. Although I'd consider some of the suggestions from other posters in this thread.

[Anduril]

Originally Posted by Justin Azevedo

hits, takeaways, giveaways and blocks are very very subjective. they fluctuate by a huge margin rink to rink as there is no official definition for what constitutes a block, hit, etc.

This is my issue with the attempt to qualify a subjective thing like grit. There are obvious issues such as home arenas skewing numbers in favor of home team and how hits, blocked shots, etc. aren't definitive of of things like grit and success.

[ricardodw]

Originally Posted by Anduril

This is my issue with the attempt to qualify a subjective thing like grit. There are obvious issues such as home arenas skewing numbers in favor of home team and how hits, blocked shots, etc. aren't definitive of of things like grit and success.

To me it passes the smell test.

I had identified Cammalleri, Hudler, Tanguay and Cervenka as small skilled guys that did not play hard... bump people over, win puck battles.

The stats and my analysis support what I see when I watch the game.


If they did not publish or highlight scoring stats I would still be able to say from watching that Iginla and Cammalleri score more goals than average. Tanguay would appear to set up more goals than other players. Keeping scoring stats just confirms this.

The RGI also provide me with a degree of comfort (hope) that Baertschi and Brodie will not be a grit liabilities as I had feared. It also helps me understand why some people are not enamoured with Glencross. He is not playing as consistently hard as he did 3-4 years ago. I have been stuck with the impression of him being a very gritty player. The RGI shows that he does indeed take too many nights off.

It also shows that with so many non-gritty players the Oilers will not be successful. I felt this was their problem but the RGI off so many important players less than 1 is a useful illustration. If Lowe and McTavish read this thread they might get a clue as to what they need to fix their team.


Feaster and co already has done a similar analysis and the Flames will not be as easy to play against this coming season.

[StrykerSteve]

Is ricardo praising Butler in this thread? O.o

I don't drink, perhaps it is time to start.

[Day Tripper]

Originally Posted by Anduril

This is my issue with the attempt to qualify a subjective thing like grit. There are obvious issues such as home arenas skewing numbers in favor of home team and how hits, blocked shots, etc. aren't definitive of of things like grit and success.

Yes. The best thing you could do is to just use road stats, which would equalize things somewhat. Not a perfect solution, but an improvement nonetheless.

Also, naming things after yourself is lame.

[kirant]

Just for kicks, I modified ricardo's formula and tried it on players who played most of last season (36+ games last season). The formula I used was:

Altered RGI (ARGI) = (Hits + Blocked Shots + Takaways - Giveaways)/(minutes per game)

This takes out the assumptions of minutes played by each player. The results of the calculation can be downloaded on this spreadsheet:

http://rapidshare.com/files/1638440078/ARGI.xls

Take it for what it is...an arbitrarily defined stat.

I also did simple cut offs by minutes played per game on subsequent sheets. The basic stats (average, median, standard deviation) are provided on each sheet.

General notes:
- Apparently it's possible to get a negative score. Ask Kovalchuk
- A score of about 3 to 8 is where the majority of players fall. Above 8 and, by ricardo's system, you'd call them grittier than most. Below 3 and you'd likely be defined as softer.
- The differences in responsibility of defence vs forwards was not considered. I deleted the position column too quickly
- Tim Jackman is way up there, making me question how valid this test really is. Players with lower minutes per game seem to do much better.
- The use of the giveaway/takeaway seems too useless in this scenario. I'm not really sure I'd call Crosby all that soft.

[theinfinitejar]

Originally Posted by kirant

- Tim Jackman is way up there, making me question how valid this test really is. Players with lower minutes per game seem to do much better.

I'd expect there to be a correlation between underwater Corsis and a high "RGI" (if the methodology was perfect). It's biased against strong possession teams and players. If you have the puck in the other teams zone all game you're not going to get any blocked shots or hits or takeaways, and you're probably going to end up giving the puck away once in a while. The Tim Jackmans of the world, meanwhile, can rack up hits and blocked shots all they want while the other team beats the pants off them. If we fixed the methodology of this stat (use away stats only, only ES numbers) I'd expect it to have a small correlation with losing.

[Textcritic]

Originally Posted by ricardodw

To me it passes the smell test.

Not really even that. There is WAAAAY more work to be done before this analysis may be considered useful.

Originally Posted by ricardodw

I had identified Cammalleri, Hudler, Tanguay and Cervenka as small skilled guys that did not play hard... bump people over, win puck battles.

The stats and my analysis support what I see when I watch the game.

Of course they do! But not necessarily because your analysis is an accurate projection of what happens in a game, much moreso because these are the things that you notice in the first place that have helped you to form your impression. Your analysis is an elaborate confirmation bias, but it is highly debatable whether it is an accurate measure of the outcome of the games.

Originally Posted by ricardodw

The RGI also provide me with a degree of comfort (hope) that Baertschi and Brodie will not be a grit liabilities as I had feared. It also helps me understand why some people are not enamoured with Glencross. He is not playing as consistently hard as he did 3-4 years ago. I have been stuck with the impression of him being a very gritty player. The RGI shows that he does indeed take too many nights off.

No, it really doesn't. It MAY show this IF you perform the same analysis for Glencross over each season of his career, which happens to result in a pattern or trend downwards. Even then, because Glencross's role on the team has changed from when he first joined the Flames, I suspect that your analysis is likely not showing a true relationship between his performance and the outcome of the games.

Originally Posted by ricardodw

It also shows that with so many non-gritty players the Oilers will not be successful. I felt this was their problem but the RGI off so many important players less than 1 is a useful illustration...

This, I tend to agree with, but not based on your statistical analysis itself. Moreso, my concern for the Oilers would be that in accordance with Justin Bourne's excellent article posted above, they appear to be a team that SHOULD be high in puck possession as opposed to puck pursuit, and yet they had abysmal results from the games played. This tells me that they are not at all successful in the type of game that they are attempting to play, and it is really difficult to see how they can or will improve on that with the collection of players that they have.

[theinfinitejar]

Originally Posted by Textcritic

Not really even that. There is WAAAAY more work to be done before this analysis may be considered useful.

You can do all the work you want, but this metric is useless from the word go.

To win hockey games you need score more goals than the other guys, and the best way to do that is to have the puck more than they do. So if you have a metric where "positive" events can only happen when the other team has the puck, how can it possibly be useful at predicting success?

[Oil Stain]

Originally Posted by ricardodw

It also shows that with so many non-gritty players the Oilers will not be successful. I felt this was their problem but the RGI off so many important players less than 1 is a useful illustration. If Lowe and McTavish read this thread they might get a clue as to what they need to fix their team.

Oilers stats guy has historically given out way more giveaways then league average so the Oilers probably can't ever do well in this stat.

Go to NHL.com and look at old Oilers-Flames games. Ones in Edmonton will usually have 2-3 times the giveaways as the ones in Calgary.