Ideas for Future Studies

[MarkGio]

Impact of Injuries

I'd like to see more data on man-games lost and it's affect on winning. Even some simple macro-level stats can start a conversation, such as 1 injured centreman reduces the chances of winning by X% league wide.

Moreover, I think a simple look at team stats after a player's injury could be a useful to examine how valuable a player is to a given club. Ultimately I'd like to know how much a team is worse off with Player-X out of the line-up.

To add fuel to the fire, I'd like to know which injuries are the most detrimental to a player's performance. Assuming the recovery time is identical, Is a broken foot worse than a broken hand at reducing point totals? Average recovery times?

This stuff and more would be gold for fantasy pools.

 

[tag0519]

I'm not sure if this exists already but what would everyone think of a stat called - "Adjusted Plus/Minus"? Seems that +/- has a lot to do with the team as well as the individual. Lets face it - Stamkos and Tavares would have a better +/- if they were on Pitt or Chi. It would be good to get an idea of a players "real" +/-.

 

[Cunneen]

I'm not sure if this exists already but what would everyone think of a stat called - "Adjusted Plus/Minus"? Seems that +/- has a lot to do with the team as well as the individual. Lets face it - Stamkos and Tavares would have a better +/- if they were on Pitt or Chi. It would be good to get an idea of a players "real" +/-.

There are several attempts to calculate adjusted plus minus. There are relatively simple versions which adjust a player's plus minus to his own team's success. This is relatively simple and still doesn't tell us much about a player's performance.

There are more in-depth, statistically driven models by statisticians that attempt to find the true value of a player. A great one is by Brian Macdonald, and it was presented at the MIT Sloan sports analytics conference. Link is below. I'd advise anyone interested in hockey analytics to take a loot at the full paper.

http://www.sloansportsconference.com/?p=2838

 

[Hardyvan123]

In James' case, it was expressed as percentage of leaguewide at-bats taken by HOFers. In any given year, it was around 10% of total at-bats that were taken by HOFers, with a massive upward tick in the 1925-34 time frame.

I think ice time would be too limiting, since we'd only be able to look at the last 13 years.

I remember this as well and James was very good about throwing ideas against the wall.

as I recall, alot of the "extra AB" from that time period came from the veterans committee inclusions to the HOF in baseball but it's been a long time since I read that book.

We had a similar bulge of Dmen from the early 70ish era in the top 60 Dman project but I think alot of it does come down to politics and engrained opinions rather than any statistical data.

It's certainly possible for certain eras to have higher or lower number of HHOFers or elite players but alot of it is subjective and can't really be quantified by pure numbers, there are other factors at work IMO.

 

[I am Lorde]

I'd love to see the average salary of players picked at each spot in the draft.

Salary is essentially what GMs and Agents estimate is the "worth" of a player. Not every player in the study would be worth exactly what he's making, but hopefully that would be averaged out over a large sample size. This would be a relatively simple way of putting a numerical value on draft picks.

I could then say "your first overall pick is worth on average a $6M player. My 9th, 10th, and 11th overall picks are each worth on average a $2M player. I'll trade you 9, 10, 11 for 1"

 

[I am Lorde]

I'd also like to see the offensive zone chopped up into "blocks" and see the average shooting percentage from each of these blocks. You could then assign a numerical value to a shot from each of these blocks.

This would reveal the difference between one player ripping 10 shots on net per game from absolutely anywhere and the player who picks his spots and takes 3 high percentage shots per game. The value of where a shot is taken would make questionable stats like "corsi" much more valuable because I want to know if my player is creating high percentage shots for and preventing high percentage shots against.

 

[Cunneen]

I'd love to see the average salary of players picked at each spot in the draft.

Salary is essentially what GMs and Agents estimate is the "worth" of a player. Not every player in the study would be worth exactly what he's making, but hopefully that would be averaged out over a large sample size. This would be a relatively simple way of putting a numerical value on draft picks.

I could then say "your first overall pick is worth on average a $6M player. My 9th, 10th, and 11th overall picks are each worth on average a $2M player. I'll trade you 9, 10, 11 for 1"

I might look into this. We could determine then at which spots in the draft the best value can be gotten, while we can also see which picks are usually over paid for.

 

[Cunneen]

I'd also like to see the offensive zone chopped up into "blocks" and see the average shooting percentage from each of these blocks. You could then assign a numerical value to a shot from each of these blocks.

This would reveal the difference between one player ripping 10 shots on net per game from absolutely anywhere and the player who picks his spots and takes 3 high percentage shots per game. The value of where a shot is taken would make questionable stats like "corsi" much more valuable because I want to know if my player is creating high percentage shots for and preventing high percentage shots against.

The only problem I would say with this is that a decent amount of research has been done that has shown that teams aren't able to generate a crazy amount of scoring chances compared to their shot differentials. Teams that have good Corsi or Fenwick or shot differential stats drive the play at even strength, and those teams are able to generate scoring chances. In the words of Vic Ferrari, "Corsi begets scoring chances, scoring chances beget goals"

links that are the research I mentioned
http://www.hockeybuzz.com/blog/Travis-Yost/The-Shot-Quality-Myth-Revisited/134/50398#.UZVHh3DR05Q

http://www.arcticicehockey.com/2011/10/25/2512376/luck-vs-shot-quality-in-shooting-percentage

http://vhockey.blogspot.com/2009/05/possession-is-everything.html

 

[Czech Your Math]

I'm not sure if this exists already but what would everyone think of a stat called - "Adjusted Plus/Minus"? Seems that +/- has a lot to do with the team as well as the individual. Lets face it - Stamkos and Tavares would have a better +/- if they were on Pitt or Chi. It would be good to get an idea of a players "real" +/-.

There is such a stat. It compares a player's ESGF/GA ratio while he's on the ice vs. that of his team while he's off the ice. It's one of the best studies done IMO:

Adjusted Even Strength Plus-Minus

 

[I am Lorde]

The only problem I would say with this is that a decent amount of research has been done that has shown that teams aren't able to generate a crazy amount of scoring chances compared to their shot differentials. Teams that have good Corsi or Fenwick or shot differential stats drive the play at even strength, and those teams are able to generate scoring chances. In the words of Vic Ferrari, "Corsi begets scoring chances, scoring chances beget goals"

links that are the research I mentioned
http://www.hockeybuzz.com/blog/Travis-Yost/The-Shot-Quality-Myth-Revisited/134/50398#.UZVHh3DR05Q

http://www.arcticicehockey.com/2011/10/25/2512376/luck-vs-shot-quality-in-shooting-percentage

http://vhockey.blogspot.com/2009/05/possession-is-everything.html

What I'm getting at is trying to put a value on scoring chances. Every shot is technically a scoring chance, but I want to see who is getting or setting up the good chances.
For example, the patented Ryan smyth slapshot as soon as he crosses the blueline. Ya he's creating shots, but he might score on 1/100 of them.
I want to know what spots on the ice players are most likely to score from, and who's shooting and setting up shots from those zones.

 

[Cunneen]

What I'm getting at is trying to put a value on scoring chances. Every shot is technically a scoring chance, but I want to see who is getting or setting up the good chances.
For example, the patented Ryan smyth slapshot as soon as he crosses the blueline. Ya he's creating shots, but he might score on 1/100 of them.
I want to know what spots on the ice players are most likely to score from, and who's shooting and setting up shots from those zones.

I think the key question is whether looking at scoring chances really adds any value. If teams are not able to consciously drive up scoring chances independent of shot differentials, then there is not much reason for people to look at scoring chances very closely. So far I haven't seen much research to show that.

 

[Pontius Palat]

Is there any quantifiable evidence to support that LH/RH defense pairings perform any better than LH/LH or RH/RH?

It seems like fans here talk more about which way defenders hold their stick than about their actual playing ability. I'd look for it myself but I'd have no idea where to start and how to go about it in the most efficient way. I'd guess a simple way would be to find certain players' +/- when playing with other players.

If anyone here wants to go for it and open a thread on the results I'd be very interested.

 

[Micklebot]

Does anyone know where sites like this get shot coordinates?

My thought is with a complete list of each teams away shots for/against coordinates (to avoid home team shot count bias) you could count the shots within the coordinates that make up the "home plate" to get a rough indicator of scoring opportunities and get a Scoring Opportunities % stat. I think it would be interesting to see how strongly that correlates with FF% or CF%, as well as win%.

Regrettably I think mining the data is likely outside my limited programming capabilities.

edit: hmm I should have read a couple posts up, looks like it's essentially been done with manually tracked scoring chances.

 

[do0glas]

Is there a way to calculate percentage of assists from something other than a shot, given the numbers that are currently out there?

we have A/60 and S/60 along with total assists/total shots. if you allow for a tolerance, could we glean any information about playmaking ability?

 

[Patman]

Is there a way to calculate percentage of assists from something other than a shot, given the numbers that are currently out there?

we have A/60 and S/60 along with total assists/total shots. if you allow for a tolerance, could we glean any information about playmaking ability?

I suppose you could go through the records and go x seconds back (at most the previous face-off) and strike those assists which come x seconds after a shot by that player. Somewhat imperfect, but I can't think of another way given "easy" data.

 

[do0glas]

I suppose you could go through the records and go x seconds back (at most the previous face-off) and strike those assists which come x seconds after a shot by that player. Somewhat imperfect, but I can't think of another way given "easy" data.

Ive been perusing the stats and for individual players ive found that there seems to be a good correlation between a low "relative" ifenwick/60 and high IPP (probably 65% and above). there are limitations to this, ifenwick can vary team to team in regards to what is low, but i considered anything under 9 in my searches. as that comes out pretty low over a single game

so for like obvious play-makers (not really using data from 12/13)

joe thornton-7.05 ifenwick/60 81.5% IPP (to me, this screams the actual ability to make a play)

martin st. louis-8.84 ifenwick/60 67.6% IPP -this one is tougher, hes taking a decent amount of shots, but hes relatively low to his top 6 counterparts.

kyle wellwood 6.97 ifenwick/60 78.1% IPP-I love wellwood, wish the sharks still had him. this guy should at least be a playmaker on a scoring third line, its hard to believe that hes not getting new contracts given his stats, advanced or otherwise.

mike ribeiro-8.274 89.4% IPP!-having that much offense off of your stick for your line, but an ifenwick under 9/60 shows impressive playmaking skills to me.

other guys that have shown the ability to make plays but "probably" generate more offense from shooting

patrick kane 11.86 ifenwick/60 75.4% IPP

brad marchand 9.163 71.4% IPP


i didnt go through every team, i was really trying to see if perceptions matched reality. and it seems they do for the most part for elite playmakers.

where it got hairy was TOI-the lower the time on ice and obviously lower point totals are something that youd need to be cognizant of. a player can generate most of his lines offense but if its only like 7 points its not worth much.

I also would go through the WOWYs and look at boosts to GF%

all the "elite" guys i put up are very good at boosting this for their linemates while not scoring a lot themselves.

would love for a way to improve this

 

[SelleckStache]

Does anyone keep track of the amount of times a goalie freezes the puck?

I guess it would get lumped into defensize zone starts, but then you'd have to take out icings.
Some goalies are better at it than others and I thing it would be an interesting way to help quantify a goalies rebound control.

 

[Bear of Bad News]

Does anyone keep track of the amount of times a goalie freezes the puck?

I guess it would get lumped into defensize zone starts, but then you'd have to take out icings.
Some goalies are better at it than others and I thing it would be an interesting way to help quantify a goalies rebound control.

I'd love to see it (and haven't). I'm not sure if the RTSS that the NHL uses tracks that sort of thing, although you could probably infer it by seeing a goalie save followed by a faceoff.

Webcrawling has never been my strong suit (all of the stats on my goalie page are hand-entered if you can believe that, but I've seen enough mistakes in official data that I like to review it personally).

 

[SelleckStache]

I'd love to see it (and haven't). I'm not sure if the RTSS that the NHL uses tracks that sort of thing, although you could probably infer it by seeing a goalie save followed by a faceoff.

Webcrawling has never been my strong suit (all of the stats on my goalie page are hand-entered if you can believe that, but I've seen enough mistakes in official data that I like to review it personally).

That's a great idea although it wouldn't include shots deflected out of the rink.. unless that is recorded too.

 

[Bear of Bad News]

That's a great idea although it wouldn't include shots deflected out of the rink.. unless that is recorded too.

That's a good point (hopefully they distinguish with something like "PUCK OUT OF PLAY", but you never know).

 

[JuJu Mobb]

Don't know if this been mentionned :

The number of times (in %) a team scores a goal when pulling the goalie.

 

[ted2019]

Is there a way to figure out true trophy winners for the pre-NHL era?

 

[Bear of Bad News]

Is there a way to figure out true trophy winners for the pre-NHL era?

In what sense? Like who would have won the Hart in pre-NHL days?

Some folks have done that with retro award balloting. I don't know how well it conforms to voting patterns of the time, and how much benefit of hindsight was used (even unwittingly).

 

[ted2019]

In what sense? Like who would have won the Hart in pre-NHL days?

Some folks have done that with retro award balloting. I don't know how well it conforms to voting patterns of the time, and how much benefit of hindsight was used (even unwittingly).

Correct. Wonder if there is some kind of formula that could be used based on winners from the NHL.

 

[thom]

Comparing elite players of different eras with the teamates they played with.Example Joe Sakic vs Stevie Y or Howie Morenz vs Jean Beliveau-who had better teamates and by how much and how did it contribute to their greatness