100% Sure this won't be the last Eberle whine

[WordsOfWisdom]

48 minutes ago, JR Ewing said:

My concern is that the tracking of stats like Hits and Shots varies greatly from arena to arena, and that's with stats for which there is general agreement. Can you imagine teams of people tracking defensive errors across the league? I think we can agree that there would be a wider range of opinion on what constitutes a mistake.

 

It would be much like baseball and what constitutes a fielding error. It is defined as "a play that should be made with ordinary effort". It has an element of subjectivity to it. There is an official scorer for each team I believe, and each one may see things slightly different.

 

Ditto for what constitutes a passed ball versus a wild pitch. Heck, even balls and strikes are subject to the opinion of an umpire. 

 

52 minutes ago, JR Ewing said:

It seems to me that both defensive player AND goalie can (and do) make mistakes on soft goals. Errors can happen because the defenseman gave up the blueline too easily, for example. If the puck carrier steps over the blue line and fires a Ryan Smyth softy by the goalie, there were still two mistakes, bad goaltending notwithstanding.

 

In a case like that, both players can be charged with an error. There is flexibility to assign 1, 2, or 3 errors on each goal (one per player). 

 

It's exactly the reverse of goals and assists in offence. We know when we see a defenceman make a perfect pass to spring a forward on a breakaway for a goal. We know that defenceman got an assist and earned it. In the same manner, we know when we see a turnover that leads to a breakaway and a goal. We know when someone was left open in front of the net or at the point for a clear shot. We can see defensive breakdowns occur just as clearly, most of the time.

 

57 minutes ago, JR Ewing said:

The numbers you want pretty much already exist: Scoring Chances Against and High Danger Scoring Chances Against. There's also Expect GF and their cousins as well. This removes bias, subjectivity and low sample size.

 

By all means, continue to track this sort of thing. It can be fun to do, and has a person really looking at the players on his team.

 

Cool.  

 

Well what I'm doing is basically an experimental project. I don't get paid for it. That guy working for the Edmonton Journal probably does. (Lucky sob lol.) I'm not familiar with what he's doing but I doubt it's exactly the same as my DE stat. This is essentially a pilot project to see if there's any interest in such a statistic for the NHL. If not, I'll cancel it. This is probably the last season I'll track it, unless I hear back from the NHL, TSN, SportsNet, Stats Inc., or someone that wants to pay me to track it for a living.

 

 

 

[Podein25]

48 minutes ago, WordsOfWisdom said:

 

It would be much like baseball and what constitutes a fielding error. It is defined as "a play that should be made with ordinary effort". It has an element of subjectivity to it. There is an official scorer for each team I believe, and each one may see things slightly different.

 

Ditto for what constitutes a passed ball versus a wild pitch. Heck, even balls and strikes are subject to the opinion of an umpire. 

 

 

In a case like that, both players can be charged with an error. There is flexibility to assign 1, 2, or 3 errors on each goal (one per player). 

 

It's exactly the reverse of goals and assists in offence. We know when we see a defenceman make a perfect pass to spring a forward on a breakaway for a goal. We know that defenceman got an assist and earned it. In the same manner, we know when we see a turnover that leads to a breakaway and a goal. We know when someone was left open in front of the net or at the point for a clear shot. We can see defensive breakdowns occur just as clearly, most of the time.

 

 

Cool.  

 

Well what I'm doing is basically an experimental project. I don't get paid for it. That guy working for the Edmonton Journal probably does. (Lucky sob lol.) I'm not familiar with what he's doing but I doubt it's exactly the same as my DE stat. This is essentially a pilot project to see if there's any interest in such a statistic for the NHL. If not, I'll cancel it. This is probably the last season I'll track it, unless I hear back from the NHL, TSN, SportsNet, Stats Inc., or someone that wants to pay me to track it for a living.

 

I've thought about your project for awhile now. 

 

Tracking individual player errors is a valuable enough exercise. But you're only tracking the errors that occur on goals. What about all the other errors committed? I'd rather know takeaways and giveaways, which is tracked right? 

 

If my expectation was that players played 82 games of mistake-free hockey, then maybe I'd care more about this particular error leading to a goal.

[JR Ewing]

50 minutes ago, Podein25 said:

 

I've thought about your project for awhile now. 

 

Tracking individual player errors is a valuable enough exercise. But you're only tracking the errors that occur on goals. What about all the other errors committed? I'd rather know takeaways and giveaways, which is tracked right? 

 

If my expectation was that players played 82 games of mistake-free hockey, then maybe I'd care more about this particular error leading to a goal.

 

This is pretty much my view as well: go for it. Have fun and follow things, but I don't think it scratches very deeply past the surface.

 

@WordsOfWisdom

 

Staples laid out his "scoring system" years ago. It was almost exactly what you describe.

 

You say that it pretty much works in the opposite way of rewarding goals and assists, yet this is also a weakness. Goals and assists tells us who the last 2 or 3 players were to touch the puck, but does an incomplete job of telling us if players are driving offense. Doug Harvey didn't pile up tons of points, yet every single viewer of a hockey game watched him completely control play while he was on the ice and a ton of offense traveled through this stick.

 

It's to that extent that penalizing the last 2 or 3 players with a defensive error doesn't really capture all of the elements in what caused a goal against. It could definitely be from completely blowing an obvious assignment next to the net. It could also be that a defensive winger completely missed his man in the neutral zone, allowing the eventual scorer to enter with speed and score several seconds later, completely away from the original picture which would have illustrated the mistake which was made. A sufficiently bad defensive player may not be remotely close to the vicinity to risk a defensive error, though not being there is one in and of itself. What of the player who wires a low percentage shot from a bad angle, only to miss high and give the opposition a two-on-one which they bury. Is that an offensive error or defensive or neither? We could probably dream up a million scenarios.

 

I also have serious doubts as to being able to find enough people who could reliably assign errors in consistently, giving us a more reliable look at what is happening. In baseball, it's an almost universally simple concept: drop a ball, throw one away, and that explains about 99% of errors. In hockey, we see wide variance, on a building-by-building basis, of how many shots and hits are counted. This is a problem to the extent that these are now viewed as Park Effects which need to be accounted for, so I can't begin to imagine how difficult it would be to add something so subjective. I don't think the pool gets any clearer if we dump a bunch of swamp water in it.

 

 

[WordsOfWisdom]

1 hour ago, Podein25 said:

 

I've thought about your project for awhile now. 

 

Tracking individual player errors is a valuable enough exercise. But you're only tracking the errors that occur on goals. What about all the other errors committed? I'd rather know takeaways and giveaways, which is tracked right? 

 

If my expectation was that players played 82 games of mistake-free hockey, then maybe I'd care more about this particular error leading to a goal.

 

I'm not sure what the status on giveaways and takeaways is. I think the game is just too fast to track those accurately. You could have several of those events occurring in a matter of seconds as players battle for puck possession. 

 

My opinion has always been this:

 

The players that commit the most errors (leading to a goal allowed) will also be the players that commit the most errors overall (assuming you could track every mistake made on the ice). It's statistical sampling basically. The errors that lead to a goal are a sample of the total population of all errors (both leading to goals and not). It would be statistically impossible for a player to lead his team in errors (that lead to a goal) and not also be at or near the top of the list for total errors. We can prove this using probability theory. If you toss a coin once, it's 50/50 whether it's heads or tails. When you toss a coin a thousand times, the odds that you're going to get 900 heads or 900 tails is less than the odds of being struck by a meteor. It just can't happen. 

 

Basically, a player can never be so unlucky as to have only his errors resulting in goals while other players commit errors that don't result in goals. The law of averages demolishes any possibility for that as the data accumulates.

 

[WordsOfWisdom]

19 minutes ago, JR Ewing said:

This is pretty much my view as well: go for it. Have fun and follow things, but I don't think it scratches very deeply past the surface.

 

Waaaaaaaaaaaaaaaaaaait! Read my post about probability theory!  

 

 

[WordsOfWisdom]

20 minutes ago, JR Ewing said:

Staples laid out his "scoring system" years ago. It was almost exactly what you describe.

 

You say that it pretty much works in the opposite way of rewarding goals and assists, yet this is also a weakness. Goals and assists tells us who the last 2 or 3 players were to touch the puck, but does an incomplete job of telling us if players are driving offense.

 

Ick. I shouldn't have said it that way. I take back my previous comment. It doesn't work anything like goals and assists. 

 

The only thing they have in common is that on a goal, there are 1, 2, or 3 points available that are distributed among players. That's where the similarity ends.

 

[WordsOfWisdom]

DEFENSIVE ERRORS - A QUICK GUIDE

 

 

HOW TO SCORE IT:

Quick FYI: A defensive error (DE) is a poor defensive play (or defensive play not made) that directly results in a goal being allowed.

Criteria/guidelines for assigning a DE:

1. For every goal allowed, charge ONE, TWO, or THREE different players on the defending team with a DE on the play.

EXCEPTION: No DE is charged on an empty net goal (when the goaltender is pulled for an extra attacker) unless it is for a penalized player (rule #3 applies).

EXCEPTION: No DE is charged on shootout goals.

NOTE: Never assign a DE for plays that do not result in a goal allowed.

NOTE: Players are responsible for "own-goals". Charge a DE to a player that shoots, tips, or redirects a puck into their own net.

2. Charge the goaltender with a DE on a "soft" goal. (Bad angle goals, long distance goals, goals due to poor rebound control, and so on.)

NOTE: The goaltender can be affected by this guideline in any situation (power play, even strength, or shorthanded).

NOTE: A bad/uncontrolled rebound that results in a goal is also considered to be a soft goal.

NOTE: The goaltender is not expected to be able to stop one-timers, break-aways, deflections, screened shots, and so on.

3. Charge the penalized player with a DE if a goal is allowed while short-handed.

NOTE: Offsetting penalties do not result in a team being shorthanded. Therefore, the penalized player cannot be charged with a DE if this occurs.

NOTE: On a 2-man advantage, charge the first player that took a penalty with a DE (whichever penalty expires first).

NOTE: If two different players on the same team are penalized at the same time (creating a 2-man advantage for the other team which they score on), the DE will be applied to both penalized players.

NOTE: Charge the penalized player with a DE on an empty net goal as well. (They prevented the extra attacker advantage that may have tied the game.)

NOTE: On a freshly expired penalty, if the player in the box cannot get back into the play, they should still be charged with a DE.

EXCEPTION: On a bench minor penalty (example: TOO MANY MEN on ice), charge the TEAM with a defensive error, but not any individual player.

4. Charge the player that creates an obvious give-away or causes an odd man rush against, from which team mates cannot recover their positioning and defend appropriately.

NOTE: The key here is whether the team can recover their defensive positioning or not. If they can, any goal scored thereafter is the result of a subsequent defensive breakdown. The original player is "off the hook" so to speak.

5. Charge the last skater involved in the play with a DE if they could have prevented the goal with proper defensive coverage but didn't.

NOTE:  A defenceman is not expected to be able to stop a 2-on-1, 3-on-1, etc. and neither is the goalie.

NOTE:  Proper defensive coverage means covering your player on the ice, having proper gap control, blocking a shot attempt if possible, not screening your own goalie, etc.

 

[Podein25]

7 minutes ago, WordsOfWisdom said:

 

I'm not sure what the status on giveaways and takeaways is. I think the game is just too fast to track those accurately. You could have several of those events occurring in a matter of seconds as players battle for puck possession. 

 

My opinion has always been this:

 

The players that commit the most errors (leading to a goal allowed) will also be the players that commit the most errors overall (assuming you could track every mistake made on the ice). It's statistical sampling basically. The errors that lead to a goal are a sample of the total population of all errors (both leading to goals and not). It would be statistically impossible for a player to lead his team in errors (that lead to a goal) and not also be at or near the top of the list for total errors. We can prove this using probability theory. If you toss a coin once, it's 50/50 whether it's heads or tails. When you toss a coin a thousand times, the odds that you're going to get 900 heads or 900 tails is less than the odds of being struck by a meteor. It just can't happen. 

 

Basically, a player can never be so unlucky as to have only his errors resulting in goals while other players commit errors that don't result in goals. The law of averages demolishes any possibility for that as the data accumulates.

 

 

I'm not a statistician, but it is not true what you say about 1,000 tosses. Eventually, you would get 900 or more heads, for example, if you did enough 1,000 tosses runs.

 

You also seem to confusing luck and it's distribution with stats. Stats are just stats, no?

[WordsOfWisdom]

2 minutes ago, Podein25 said:

 

I'm not a statistician, but it is not true what you say about 1,000 tosses. Eventually, you would get 900 or more heads, for example, if you did enough 1,000 tosses runs.

 

You also seem to confusing luck and it's distribution with stats. Stats are just stats, no?

 

Stats are just stats, but they all accumulate according to statistical probabilities.

 

If you tossed a coin 1000 times, the most likely result that you would get is 500 heads and 500 tails. Although you probably wouldn't get 500/500 exactly, it would be close. For it to be 900/100 is so unlikely as to be virtually impossible. It's still "possible" in theory but it would never occur in practice. Hope that makes sense.  

[Podein25]

21 hours ago, WordsOfWisdom said:

 

Stats are just stats, but they all accumulate according to statistical probabilities.

 

If you tossed a coin 1000 times, the most likely result that you would get is 500 heads and 500 tails. Although you probably wouldn't get 500/500 exactly, it would be close. For it to be 900/100 is so unlikely as to be virtually impossible. It's still "possible" in theory but it would never occur in practice. Hope that makes sense.  

 

Never is a long time you know.

 

I'm now bowing out of this convo and return to subjects with which I'm more familiar. Like sheep.

[WordsOfWisdom]

57 minutes ago, Podein25 said:

 

Never is a long time you know.

 

I'm now bowing out of this convo and return to subjects with which I'm more familiar. Like sheep.

 

LOL.   No prob. I never like to sound like I'm being argumentative or anything.

 

Statistics is one of the most difficult subjects I ever took in school. 

 

[Podein25]

45 minutes ago, WordsOfWisdom said:

 

LOL.   No prob. I never like to sound like I'm being argumentative or anything.

 

Statistics is one of the most difficult subjects I ever took in school. 

 

 

Your went to school?!

[WordsOfWisdom]

3 hours ago, Podein25 said:

 

Your went to school?!

 

Yeah. Fat lotta good it did though.

 

I was always told that getting an education would lead to a higher paying job. LOLOLOLOLOL yeah right. I guess that may have been the case in the days before everyone and their dog started getting Masters degrees.