Ongoing Stats and Analytical Discussion Thread: Battle of the Defense

[Feed Me A Stray Cat]

So many discussions veer off into the realm of why stats are wrong/incomplete or why you should use stats and what not. Figured it would be good to eventually move them into this thread. Also could serve as a good resource for people wanting to learn more about it.

Below are some helpful links:

http://stats.hockeyanalysis.com/index.php
Great website that has player and team data going back through 2007 on a variety of "advanced" stats, such as Fenwick, Corsi, Shots, Goals, Per Minute Production, etc. across all manpower and score situations.

http://corsica.hockey/
Similar concept to the previous link, but offers some additional information such as scoring chance data, expected goals data and more.

http://www.hockey-reference.com/play-index/psl_finder.cgi
Hockey reference play index allows you to parse out leaders in various statistics through specific time frames and subject to various criteria. For example, a list of players with the best PPG in the NHL from 2007 to 2014 with a minimum of 400 games played?

http://www.nhl.com/stats/player?agg...r=gamesPlayed,gte,1&sort=points,goals,assists
NHL.com's stat database has become more robust. You can search certain advanced stats, and also provide games played requirements as well as set start and end points for the time frames you want to search. For instance, who has the best faceoff percentage in the NHL since January 5th with at least 10 games played (not that you would ever want or need that specific stat)?

Some seminal articles on stats:

http://www.arcticicehockey.com/2013...nd-misconceptions-of-advance-hockey-analytics

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

http://nhlnumbers.com/2012/5/11/the...stent-goaltenders-vs-inconsistent-goaltenders

http://www.arcticicehockey.com/2011...offensive-zone-faceoff-even-strength-vs-power

http://www.broadstreethockey.com/2013/6/13/4427522/shooting-percentage-regression

http://www.arcticicehockey.com/2013/7/26/4513170/winnipeg-jets-nhl-stats-probabilities-luck

http://www.arcticicehockey.com/2010/9/9/1677538/jody-shelley-vs-ilya-kovalchuk

Click to read more...

 

[Feed Me A Stray Cat]

To continue the discussion from the team thread:

Take Shayne Gostibehere, who had an amazing rookie season. Obviously a very talented defenseman, but his overall shooting percentage last year was 11.2% and his ES shooting percentage was 9.6%.

For some perspective:

http://www.nhl.com/stats/player?agg...&filter=gamesPlayed,gte,200&sort=shootingPctg
The best overall defenseman shooting percentage over the previous four years, min. 200 games, was Nick Holden and Shea Weber at 9.4%. NHL median of 5.0%.

http://stats.hockeyanalysis.com/rat...id=0&type=individual&sort=ishpct&sortdir=DESC
The best defenseman shooting percentage in the NHL at ES over the last four years is Brent Burns at 8.1%. NHL median is around 4.2%.

By understanding that context, you would have understood that Gostisbehere had a very high likelihood of seeing a regression in his shooting percentage this season.

Time and time again this context is brought up, and disbelievers claim that this team or player is different, and will continue to outperform everyone else. But it rarely if ever happens.

Gostisbehere is shooting at 3.3% overall and 2.7% at even strength this season. Meanwhile, he's getting more shots on net and his possession numbers are better. Is he actually playing worse offensively or were his numbers last year completely unsustainable? Maybe a bit of both. Teams have also probably adjusted. But a sizable drop off in production was pretty easy to predict.

 

[devilsblood]

Non linear results is different then luck driven results.

 

[devilsblood]

Is this an analytics thread, or a thread about the role of luck in sports?

 

[Feed Me A Stray Cat]

And yet teams have begun to implement shifts to increase their "luck". Or a really slick fielder making plays, or a really great outfielder tracking down balls, at times sacrificing hitting in order make more plays in the field. Or vice versa sacrificing fielding in order to get a good hitter into the game.

And of coarse that can all be for naught if it is a big time pitcher and he is dealing.


Or a guy can go steroids and just hit it beyond the wall where no fielders exist.

We also have data on spray charts, knowing that hitters who go center and opposite field have higher batting averages. We also know how much players are shifted against. We know someone's batting average against the shift. We can account for all those things.

The bottom line is that in baseball, a sport where analytics are much more advanced than hockey, we have gotten to the point where the existence of luck can't be denied.

It is a bit harder to definitively prove out luck in hockey. We can only really do it on the extremes - we know Rikard Rakell's shooting percentage is going to come down in the near future. We know that Patrice Bergeron's will go up. But we don't know much about the player's in the middle.

 

[AfroThunder396]

Luck can't be ignored, specifically when looking at small data sets. But as you acquire more data points the trends will normalize, this is one of the pillars of data analytics.

The problem is that an 82 season, relatively, is not a large sample size. When guys have careers lasting 800-1200, one season of success/failure isn't convincing. It's a snapshot into 5-10% of a career, which doesn't tell you a whole lot. One great/awful season statistically could easily be explained as "just a bad year" and not necessarily a drop in quality of play.

The best way to use analytics is to look at data sets as big as you can, so you have many many points from which your "luck" driven points are filtered or at least normalized. For example, I can say with supreme confidence that Cory Schneider is an elite goaltender, even though his stats from this current season are not impressive. I'm more inclined to believe the 270 games of exceptional goaltending (86% of his career) are a a more accurate gauge of his talent than this 14% stretch of disappointing play.

Perhaps "luck" isn't an appropriate term. In my field we call it "drift". Luck implies a fortunate outcome, when in reality a spontaneous and random event just happens independent of the outcome. And in reality, most drift events are neutral. Drift happens everywhere, all the time. The only real defense you have against it is to collect as many data points as you can and hope the noise gets filtered out.

 

[Edmonton East]

I predict a statistically significant uptick in infractions within this thread relative to other threads on this board.

But yea, I thought about making this a few weeks back. Had to be done I think.

 

[devilsblood]

We also have data on spray charts, knowing that hitters who go center and opposite field have higher batting averages. We also know how much players are shifted against. We know someone's batting average against the shift. We can account for all those things.

The bottom line is that in baseball, a sport where analytics are much more advanced than hockey, we have gotten to the point where the existence of luck can't be denied.

It is a bit harder to definitively prove out luck in hockey. We can only really do it on the extremes - we know Rikard Rakell's shooting percentage is going to come down in the near future. We know that Patrice Bergeron's will go up. But we don't know much about the player's in the middle.

Ha ha, Short version: Is it lucky that he did not strike out?

 

[devilsblood]

We also have data on spray charts, knowing that hitters who go center and opposite field have higher batting averages. We also know how much players are shifted against. We know someone's batting average against the shift. We can account for all those things.

The bottom line is that in baseball, a sport where analytics are much more advanced than hockey, we have gotten to the point where the existence of luck can't be denied.

It is a bit harder to definitively prove out luck in hockey. We can only really do it on the extremes - we know Rikard Rakell's shooting percentage is going to come down in the near future. We know that Patrice Bergeron's will go up. But we don't know much about the player's in the middle.

Long version.

Aight the bloop base hit.

Pitcher is trying to put the ball past the hitter, but is not able to, hitter is trying to hit it out of the ball park, but is not able to, but the pitcher was able to keep the hitter from ripping the ball, while the hitter was able to put the ball into the field of play.


The defense is aligned in such a way so as to make an out, but the defense cannot cover the entirety of the field. There is knowledge, that even though a hitter may not make great contact, the ball may land in a spot where the defense can not make a play. They could place all 9 guys 150 ft from home plate and such bloops would almost never fall, but they would then be susceptible to much bigger plays.

So there are numerous competing aspects going on here, and in this case the hitter won a small battle, he put the ball into the field of play, an area that we all agree the defense can not completely cover. Now if he swung and missed, this could not have happened.

Now is that luck? The ability to hit a ball the pitcher doesn't want you to hit, and put it into an area in which we know the opposing team can not cover the entirety of?

Imo it's not. It's the nature of competition, It's the nature of the game in which a player need not win every battle big, it's the nature of the game which forces a team to allow small defeats in order to avert larger ones.

Is that semantics, calling it the nature of the game, or the nature of competition as opposed to luck? Again, imo it is not, because calling it luck is just saying, things happen we have no control over. Where as calling it the nature of competition is saying there are numerous intentional efforts factoring into such a play.

 

[devilsblood]

Luck can't be ignored, specifically when looking at small data sets. But as you acquire more data points the trends will normalize, this is one of the pillars of data analytics.

The problem is that an 82 season, relatively, is not a large sample size. When guys have careers lasting 800-1200, one season of success/failure isn't convincing. It's a snapshot into 5-10% of a career, which doesn't tell you a whole lot. One great/awful season statistically could easily be explained as "just a bad year" and not necessarily a drop in quality of play.

The best way to use analytics is to look at data sets as big as you can, so you have many many points from which your "luck" driven points are filtered or at least normalized. For example, I can say with supreme confidence that Cory Schneider is an elite goaltender, even though his stats from this current season are not impressive. I'm more inclined to believe the 270 games of exceptional goaltending (86% of his career) are a a more accurate gauge of his talent than this 14% stretch of disappointing play.

Perhaps "luck" isn't an appropriate term. In my field we call it "drift". Luck implies a fortunate outcome, when in reality a spontaneous and random event just happens independent of the outcome. And in reality, most drift events are neutral. Drift happens everywhere, all the time. The only real defense you have against it is to collect as many data points as you can and hope the noise gets filtered out.

This I think is my argument as well.

For example, wobbly pass to d-man on point, puck bounces over his stick, opposing player takes off on breakaway and scores. Is that luck? Again, we are in competition here. One guy makes a bad pass, another guy is not good enough to handle the pass, sure you benefit from weaker competition. But that is not luck.

Or let's say, guy stationed in front of goal. Intent here is to create traffic, shot from point, point just goes in off players leg and into net. Was the player really trying to deflect the shot off his leg and into a certain area of the net? No he wasn't. But he was trying to create traffic, he was aware the possibilty of a deflection would make it more difficult for the goalie to stop. So the result was a vague rendition of the intent, but not the intent precisely. Is that luck? No.

Luck is rimming the puck around the glass it going off a stanchion and into the net. Now that is luck.

 

[Triumph]

I've called it chance before and gotten yelled at. I don't know why this word causes such a problem. Playing major league sports at any level takes an incredible amount of athleticism, learned skill, and dedication. There's also a large degree of fortune involved with who wins and loses, especially in hockey where there's only an average of 5 and a half scoring plays in a game.

 

[Triumph]

This I think is my argument as well.

For example, wobbly pass to d-man on point, puck bounces over his stick, opposing player takes off on breakaway and scores. Is that luck? Again, we are in competition here. One guy makes a bad pass, another guy is not good enough to handle the pass, sure you benefit from weaker competition. But that is not luck.

Or let's say, guy stationed in front of goal. Intent here is to create traffic, shot from point, point just goes in off players leg and into net. Was the player really trying to deflect the shot off his leg and into a certain area of the net? No he wasn't. But he was trying to create traffic, he was aware the possibilty of a deflection would make it more difficult for the goalie to stop. So the result was a vague rendition of the intent, but not the intent precisely. Is that luck? No.

Luck is rimming the puck around the glass it going off a stanchion and into the net. Now that is luck.

Parsing what 'is' and 'isn't' luck in a game with as many variables as hockey is a fool's errand. If you can't see that there are gigantic chance elements in a hockey game, I can't explain it to you. It's something people either see or they don't see.

 

[devilsblood]

I've called it chance before and gotten yelled at. I don't know why this word causes such a problem. Playing major league sports at any level takes an incredible amount of athleticism, learned skill, and dedication. There's also a large degree of fortune involved with who wins and loses, especially in hockey where there's only an average of 5 and a half scoring plays in a game.

I think we all agree they make a ton of $$$$, but what does that have to do with the discussion?

 

[SteveCangialosi123]

Long version.

Aight the bloop base hit.

Pitcher is trying to put the ball past the hitter, but is not able to, hitter is trying to hit it out of the ball park, but is not able to, but the pitcher was able to keep the hitter from ripping the ball, while the hitter was able to put the ball into the field of play.


The defense is aligned in such a way so as to make an out, but the defense cannot cover the entirety of the field. There is knowledge, that even though a hitter may not make great contact, the ball may land in a spot where the defense can not make a play. They could place all 9 guys 150 ft from home plate and such bloops would almost never fall, but they would then be susceptible to much bigger plays.

So there are numerous competing aspects going on here, and in this case the hitter won a small battle, he put the ball into the field of play, an area that we all agree the defense can not completely cover. Now if he swung and missed, this could not have happened.

Now is that luck? The ability to hit a ball the pitcher doesn't want you to hit, and put it into an area in which we know the opposing team can not cover the entirety of?

Imo it's not. It's the nature of competition, It's the nature of the game in which a player need not win every battle big, it's the nature of the game which forces a team to allow small defeats in order to avert larger ones.

Is that semantics, calling it the nature of the game, or the nature of competition as opposed to luck? Again, imo it is not, because calling it luck is just saying, things happen we have no control over. Where as calling it the nature of competition is saying there are numerous intentional efforts factoring into such a play.

Bloop base hits are absolutely luck many many times. No question about it. Your premise that the player is trying to hit it out the park isn't always true too. With Chris Davis, sure, he's trying to hit one out every at bat regardless of the count. Ichiro is almost never taking that approach and is always trying to put it in play. Also pitchers aren't always going for a swing and miss. Many pitchers aren't strikeout pitchers and are trying to get soft contact and keep it on the ground. Someone like Chien-Ming Wang or another sinker baller doesn't mind if the batter puts it in play. They're banking on soft ground balls which are more often than not outs. If they squeak through, many times that's lucky.

If a player blasts a line drive right back to the pitcher and it's caught, that is bad luck. If a player nubs one off the end of the bat and it barely loops over the third baseman's head, that is good luck. Consistently barrelling up the ball and hitting it hard will yield better results over time than hitting it off the end of the bat.

Now it's true that it's a skill to put the ball in play. Look at Ichiro's career K rate compared to Chris Davis. It's certainly a skill to get the bat on the ball with 2 strikes. This doesn't change the fact that a softly hit ball barely dropping in isn't lucky. Bat control and controlling where you hit the ball is certainly a thing, but it's still generally lucky to get a bloop. No one tries to hit it off the end of the bat or get jammed.

 

[devilsblood]

No it is the nature of the game. Not all well hit balls are base hits. Thus it is not unlucky if a well hit ball is caught. The converse is also true.

 

[Triumph]

I think we all agree they make a ton of $$$$, but what does that have to do with the discussion?

If you're going to make stupid puns, the mods may as well just boot you from this now

 

[SteveCangialosi123]

No it is the nature of the game. Not all well hit balls are base hits. Thus it is not unlucky if a well hit ball is caught. The converse is also true.

Actually, it's both. It's the nature of the game of baseball that luck plays a major role. If a pitcher perfectly executes his pitch and a guy nubs one that goes 4 feet and he legs out a single, that is absolutely 100% a lucky outcome for the batter. The pitcher made his pitch exactly like he wanted to and the batter didn't get good wood on it yet there was a bad outcome for the pitcher. If a batter consistently hits it off the end of the bat, he'd be out of the league in no time. If a pitcher consistently gets guys to hit off the end of the bat, he's an ace.

 

[devilsblood]

Parsing what 'is' and 'isn't' luck in a game with as many variables as hockey is a fool's errand. If you can't see that there are gigantic chance elements in a hockey game, I can't explain it to you. It's something people either see or they don't see.

Yeah man we see the game way different for sure.

 

[devilsblood]

Actually, it's both. It's the nature of the game of baseball that luck plays a major role. If a pitcher perfectly executes his pitch and a guy nubs one that goes 4 feet and he legs out a single, that is absolutely 100% a lucky outcome for the batter. The pitcher made his pitch exactly like he wanted to and the batter didn't get good wood on it yet there was a bad outcome for the pitcher. If a batter consistently hits it off the end of the bat, he'd be out of the league in no time. If a pitcher consistently gets guys to hit off the end of the bat, he's an ace.

Does the catcher have legs in this scenario? Perhaps doesn't have a throwing arm?

 

[SteveCangialosi123]

Does the catcher have legs in this scenario? Perhaps doesn't have a throwing arm?

There are times where a ball is placed where no one can get to it in time. I mean you aren't even taking a defensible position here...There's absolutely no question luck plays a role in baseball on a game to game basis. And that's true of every sport.

 

[devilsblood]

If you're going to make stupid puns, the mods may as well just boot you from this now

Ah c'mon man, I was completely joshing, did I need to throw a smiley face on there smooth things over.

 

[devilsblood]

There are times where a ball is placed where no one can get to it in time. I mean you aren't even taking a defensible position here...There's absolutely no question luck plays a role in baseball on a game to game basis. And that's true of every sport.

I already answered a similar scenario above in a long winded post.

 

[SteveCangialosi123]

I already answered a similar scenario above in a long winded post.

You answered it incorrectly tho.

 

[devilsblood]

You answered it incorrectly tho.

If it's a slow runner and if there is a good fielder anywhere amongst the catcher pitcher or third baseman, this base hit does not happen.

If it's a fast runner and a bunch of guys with no range in the field, that dribbler has a much larger area into which a hit will result.

 

[Triumph]

Ah c'mon man, I was completely joshing, did I need to throw a smiley face on there smooth things over.

My post was about how poor word choice has sidetracked discussions about this topic. You respond, using my new word choice as an opportunity for a pun, full stop, nothing else to your post. But it did get me to stop discussing this topic with you, so I guess I have to thank you for hollowing out a nice divot with a pillow in that brick wall.