Interesting advanced stats on the Flames' Dman passing

[Huntingwhale]

I got angry when the author showed his charts on Kris Russell and said that it should fully put to rest any notion that he is a capable top-four defenseman.Why, because of his stupid charts that mean #### all for someone who watches him play?

And I finally stopped reading when he asked the question Are Wideman's offensive characteristics worth the risk of top-four usage? Gee, I dunno. He only set a career high in points and was 4th overall in Dman scoring. Is 4th in scoring worth top 4 minutes? Let's ask Hartley, the Jack Adams' guy, what he thinks.

All the more reason to hate advanced stats. Sad article. I can't take anyone seriously who tosses out charts, backslashes and numbers without actually watching the players do their thing.

[DJones]

Originally Posted by Huntingwhale

I got angry when the author showed his charts on Kris Russell and said that it should fully put to rest any notion that he is a capable top-four defenseman.Why, because of his stupid charts that mean #### all for someone who watches him play?

And I finally stopped reading when he asked the question Are Wideman's offensive characteristics worth the risk of top-four usage? Gee, I dunno. He only set a career high in points and was 4th overall in Dman scoring. Is 4th in scoring worth top 4 minutes? Let's ask Hartley, the Jack Adams' guy, what he thinks.

All the more reason to hate advanced stats. Sad article. I can't take anyone seriously who tosses out charts, backslashes and numbers without actually watching the players do their thing.

Agreed. I like their website as they put lots of effort into it even if I don't agree with but they are single handily making me completely disrespect advanced stats guys.

[the2bears]

Originally Posted by ricardodw

The whole commentary on Wideman and Russell seems like a cry for help.... Stop us doing this we are wasting our time and are confused beyond redemption.

Russel and Wideman were the best #1 pairing in the league once Gio got hurt

Wideman 19 game 19 pts +11 27 minutes a game.
Russell 19 games 14 pts +8 26 minutes a game

Flames 12-3-4 28 out of a possible 38 pts a 120 pt season pace only about 1 time every 5 years does any team go over 120 pts a season.

There are no easy match ups when you are playing every second shift.

Can they do that for a full season? I don't think so. Weber and Sutter playing together as 26 year olds in 2011-12 could not match what Wideman and Russell did the last 19 games of 2014-15.

Any statistics that analyzes their play last year and come up with the conclusion that they are not even top 4 d-men is not WRONG , they are terribly, horribly and irredeemably WRONG.

This is actually a very good post.

[mikephoen]

One thing that drives me insane with 'advanced' stats is the notion that taking a shot and having it blocked is a good thing. Where did this come from? If your shot is blocked that is a giveaway! It ends your possession, it doesn't improve it!

How often does a shot that is blocked result in a scoring chance for the shooting team? Almost never? At best it is like a dump in and leads to a puck battle for the offensive team. More likely, it is just a giveaway. And sometimes it's a giveaway where the shooter has taken himself out of defensive position, which leads to an odd man rush the other way.

This is what happened a ton when Russell and Wideman were on the ice last year. Russell blocks the shot. Wideman or the centre recover the puck. Quick first pass to a winger and the Flames are out of the zone and on the attack. Then this dude checks his spread sheet and says 'Another bad shift for Wideman and Russell! Derp!'

Start scoring shot blocks as take always and getting a shot blocked as a giveaway and maybe some of these 'advanced' stats will begin to pass the eyeball test.

[Street Pharmacist]

Originally Posted by mikephoen

One thing that drives me insane with 'advanced' stats is the notion that taking a shot and having it blocked is a good thing. Where did this come from? If your shot is blocked that is a giveaway! It ends your possession, it doesn't improve it!

How often does a shot that is blocked result in a scoring chance for the shooting team? Almost never? At best it is like a dump in and leads to a puck battle for the offensive team. More likely, it is just a giveaway. And sometimes it's a giveaway where the shooter has taken himself out of defensive position, which leads to an odd man rush the other way.

This is what happened a ton when Russell and Wideman were on the ice last year. Russell blocks the shot. Wideman or the centre recover the puck. Quick first pass to a winger and the Flames are out of the zone and on the attack. Then this dude checks his spread sheet and says 'Another bad shift for Wideman and Russell! Derp!'

Start scoring shot blocks as take always and getting a shot blocked as a giveaway and maybe some of these 'advanced' stats will begin to pass the eyeball test.

If you block a shot, you're less likely to get it back. The more a team blocks shots, the more shots the other team gets and the fewer they get. It's a pretty strong inverse correlation.

I get you feel like that's the case, but there's a reason Wideman and Russell were on the ice for fat more shots against than for. You can try to argue that a good strategy, but you can't say blocking shots helps you get the puck. There's lots of evidence to the contrary

[mikephoen]

Originally Posted by Street Pharmacist

If you block a shot, you're less likely to get it back. The more a team blocks shots, the more shots the other team gets and the fewer they get. It's a pretty strong inverse correlation.

I get you feel like that's the case, but there's a reason Wideman and Russell were on the ice for fat more shots against than for. You can try to argue that a good strategy, but you can't say blocking shots helps you get the puck. There's lots of evidence to the contrary

That's not exactly what I'm saying. I'm saying that taking a bad shot isn't a good thing. Getting your shot blocked is not a positive, but corsi scores it like is. Taking a bad angle shot, missing the net, and having it ring around the boards and out of the zone is scored by corsi guys as a positive. That makes zero sense. That is a bad play for the offensive team.

[Frank MetaMusil]

Originally Posted by mikephoen

That's not exactly what I'm saying. I'm saying that taking a bad shot isn't a good thing. Getting your shot blocked is not a positive, but corsi scores it like is. Taking a bad angle shot, missing the net, and having it ring around the boards and out of the zone is scored by corsi guys as a positive. That makes zero sense. That is a bad play for the offensive team.

It's part of the Flames strategy. Less pucks to the net, less rebounds, less traffic in front of the goalie.

The Secret to Russell's shot blocking record

Usually when you block a shot and you’re in good position, you can transition the puck going the other way, so it benefits the offence as well.

[Oil Stain]

Originally Posted by mikephoen

That's not exactly what I'm saying. I'm saying that taking a bad shot isn't a good thing. Getting your shot blocked is not a positive, but corsi scores it like is. Taking a bad angle shot, missing the net, and having it ring around the boards and out of the zone is scored by corsi guys as a positive. That makes zero sense. That is a bad play for the offensive team.

If you do that consistently then the other team will get more shoots and you will be a poor corsi player.

Having the puck in your opponents end, and being able to take a shot is better than not having the puck in your opponents end and not being able to take a shot.

Even if it does get blocked.

[Cheerio]

Originally Posted by mikephoen

That's not exactly what I'm saying. I'm saying that taking a bad shot isn't a good thing. Getting your shot blocked is not a positive, but corsi scores it like is. Taking a bad angle shot, missing the net, and having it ring around the boards and out of the zone is scored by corsi guys as a positive. That makes zero sense. That is a bad play for the offensive team.

I think you're not really understanding the point of corsi/fenwick etc. The reason these stats were invented was because there were 0 ways to track possession in the NHL. If you watch a soccer game, where ball possession is tracked very well, you'll notice that possession correlates to winning. Corsi and fenwick are the best indicators of possession that exist for hockey, if you're shooting the puck, you have possession of it (obviously) and if the puck is being shot at your net, you don't. So corsi doesn't treat a bad angle shot, or a blocked shot as a good thing, it's treating it as a single data point where team X had the puck, and team Y didn't.

[GranteedEV]

Originally Posted by mikephoen

One thing that drives me insane with 'advanced' stats is the notion that taking a shot and having it blocked is a good thing. Where did this come from? If your shot is blocked that is a giveaway! It ends your possession, it doesn't improve it!

There's an extent to which blocking shots is good. Good players still block shots... Mark Giordano, Roman Josi, Hjalmarsson. But when you're among league leaders in the category it means you were among the league leaders in not having the puck. Not only that, but you failed to have the puck while the opponent is in your zone. See the problem? Now in some cases, like with Lance Bouma and Mikael Backlund, that's often because you're being sent out there for those situations. There's a context to it.

But Russell? I love the guy and hope he's right here for our Cup. BUT for his entire career, the guy has possession stats under 50% BUT if you look at his deployment, he starts more than 55% of his shifts in the offensive zone. That's a bad sign. It's a sign that tells you "don't force him to play against highly skilled top 6 lines because he's outmatched". And yes, he is outmatched against top lines. He was outmatched in the Vancouver series against the Sedins, if that was Daniel Sedin in his prime on those scoring opportunities we would have been swept but luckily they had zero finish. He was outmatched in the Anaheim series. He was a -8 and it wasn't just Getzlaf and Perry that were having their way with him but it was even Matt Belesky's eyes lit up playing us.

That's a problem. A problem you can hide underneath Hamilton, Giordano, and Brodie? Maybe, but I'd prefer it to not be a problem at all, but rather for Russell to be an advantage for the team and for that to happen, a bottom pairing role just makes more sense, with the confidence that if you need him to step into the middle pairing due to injury, he won't ever be a disaster. It's just a better plan than "hey, we're going to spend about 56% of the game without the puck diving to block shots and scrambling to clear the zone without icing it. All season long."

It's a better plan because

1) It puts the puck on the sticks of our forwards instead of on the sticks of the other team's forwards. That's more opportunities to score and less opportunities for the other team to score. Then we don't need to pray for a 4 on 4, a power play, or pulling the goalie to get scoring chances every game (last year's power play goals differential was the reason we were a great team, we were the tenth worst 5 on 5 team at outscoring the opposition).

2) It gives Giordano and Brodie an easier time if they can spend less time starting in their own zone. This pairing had tougher deployment than pretty much every other elite pairing in the NHL. It's not fair to them and honestly it's a waste of just how good they are for them to spend so much time defending each game. If Gio and Brodie have an easier time, then their already elite play just becomes that much more dominant because now we're the team hemming other teams in. How awesome was that sequence in Game 1 vs VAN where Russell scored the game winner? Now imagine if we got a sequence or two like that every game, instead of once a season. It also means Gio, Johnny, Monny, Hudler, Brodie etc can use the extra energy to make offense.

Again, we were not a terrible team last year like people misusing possession stats will tell you. But that doesn't mean we should be satisfied with the status quo of being a terrible possession team. Russell being moved to a bottom pairing role, hopefully with a possession-driving strong-skating partner (I really hope Nakladal or Morrisson can be that guy, because Wideman isn't) and getting Hamilton a partner who can be steady but drive possession positively (I really hope Wotherspoon can jump in and be that guy, because Russell isn't) will be a step in the right direction. Getting Engelland off the team is something that must be done if you want to get away from that status quo, because he does not do anything positive.

Russell has shown he can have a corsi over 50% but the only time this happened was in St. Louis. There he was used on the bottom pairing limited to around 16-17 minutes per game. That's what needs to happen for him here to best utilize him. Don't look at the stats as damning, look at them as illuminating.

Though the only thing they illuminate about Deryk Engelland, is that his performance is damning.

[jayswin]

Originally Posted by Street Pharmacist

Well there is. If your boxing strategy is to dodge punches all match, you'll lose every time

I remember this was one thing that bugged me about Regehr back in the day, but I didn't really say anything back then because most CP'ers were heavily bought into the Sutter "Defense and hard checking first, puck moving and skill second" mindset.

But he used to always make the safe play, which for his skill set was usually the best play, but still it was frustrating seeing him make a really good, tough checking play in our zone to recover the puck, then came the old div 6 bantam dump up high off the glass, followed by the opposing team rushing back in our zone six seconds after Regehr worked his ass off to get the puck in our possession.

[mikephoen]

Originally Posted by Cheerio

I think you're not really understanding the point of corsi/fenwick etc. The reason these stats were invented was because there were 0 ways to track possession in the NHL. If you watch a soccer game, where ball possession is tracked very well, you'll notice that possession correlates to winning. Corsi and fenwick are the best indicators of possession that exist for hockey, if you're shooting the puck, you have possession of it (obviously) and if the puck is being shot at your net, you don't. So corsi doesn't treat a bad angle shot, or a blocked shot as a good thing, it's treating it as a single data point where team X had the puck, and team Y didn't.

I understand what the proponents of corsi and fenwick want them to do. I don't agree that they accomplish that.

Taking a bad shot is a bad play. It might indicate that you used to have the puck, but it doesn't indicate that you made the right play with it.

You say Corsi and Fenwick are the best indicators of possession that exist, but I say they aren't good enough. It won't be long and we'll have real possession data once the players and puck are all chipped up, and my contention is that we'll find out that corsi and fenwick weren't really tracking much of anything useful.

[DJones]

Originally Posted by Cheerio

I think you're not really understanding the point of corsi/fenwick etc. The reason these stats were invented was because there were 0 ways to track possession in the NHL. If you watch a soccer game, where ball possession is tracked very well, you'll notice that possession correlates to winning. Corsi and fenwick are the best indicators of possession that exist for hockey, if you're shooting the puck, you have possession of it (obviously) and if the puck is being shot at your net, you don't. So corsi doesn't treat a bad angle shot, or a blocked shot as a good thing, it's treating it as a single data point where team X had the puck, and team Y didn't.

I never understood why shot attempts were chosen to be the indicators of choice for hockey. The correlation is #### and there are so many little problems with it that it becomes meaningless. Time in the offensive zone and scoring chances seem like it would make way more sense.

I read that websites stat summaries for every game as I can see the value of stats in other sports but just can't see it in hockey besides a quick judgement of a player you don't have time to watch. But about 2/5 games they completely disregard Corsi and Fenwick as meaningless and not a representation of the game at all. When it doesn't work they go into score effects and finally shot quality if it still doesn't fit the eye test. So two out of five games the stats don't fit how the game was played but then a month later the only metric they use is Corsi and Fenwick in judging the teams performance. I've literally seen them write a 1000 words on why even though we got a low corsi one game we demolished the other team and it was score effects but then the next month they point to that game as a troubling sign of our bad possession numbers.

There is so many logical fallacies involved in it that it is insane;

There is no reason why a team should have an average PTO. The average PTO is a completely meaningless number. SH% and Save % have no correlation whatsoever with eachother. The 100 or 101 average is just the average amount of a scatter point of numbers. There is no driving force pulling it towards that center.

Same with team shooting percentage, the average 9.0% is completely meaningless. Top line forwards are upwards of 13-15%, defense are 5% and everyone else is somewhere in between. The assumption that every team is taking the same distribution of shots between the values is ridiculous.

These guys go deep into advanced stats and some of them are very interesting but then they go to the next game and all they can bring forward is the basic stats that they have just spent their time disproving. Just doesn't make sense to me.

[GranteedEV]

Originally Posted by mikephoen

You say Corsi and Fenwick are the best indicators of possession that exist, but I say they aren't good enough.

The best indicator of possession that exists is measuring possession time. With the chipped pucks the NHL is introducing that's probably going to be phased in. Regardless, it doesn't matter if Fenwick etc are not perfect. They're highly correlated with winning:

5vs5 Fenwick Percentage, NHL Stanley Cup Champions:

2015 Hawks - 52.7 (7th best)
2014 Kings - 56.1 (Best)
2013 Hawks - 54.9 (3rd best)
2012 Kings - 53.6 (4th best)
2011 Bruins - 50.1 (16th best)
2010 Hawks - 57.8 (Best)
2009 Penguins Post-Thierren Firing - 53.4 (6th best)
2008 Red Wings - 59.0 (Best)

As far back as the stats exist the only team to win the cup that wasn't very good in the Fenwick stat was a Bruins team that most feel only won because Tim Thomas stole it for them in what might be the best goaltending performance ever.

5vs5 Fenwick Percentage, 2015 Flames - 45.8 (28th best)

I'd love to believe we can defy logic, but we're not yet even close to that Boston team that got by on goaltending. Does it really matter if it's a 1:1 ratio of corsi to possession? The real takeaway is that Cup champs are good at it, and right now the Flames are not. The Flames also were not cup champs to disprove that.

ONLY getting better at just Corsi won't make the Flames a better team, but we need to keep adding good hockey players and cutting bait with bad ones without getting emotionally attached. Brad T's already shown a willingness to do so with Frolik and Hamilton... he clearly didn't think Colborne/Jones or Wideman were "good enough" in their roles last year.

[jayswin]

Originally Posted by mikephoen

I understand what the proponents of corsi and fenwick want them to do. I don't agree that they accomplish that.

Taking a bad shot is a bad play. It might indicate that you used to have the puck, but it doesn't indicate that you made the right play with it.

You say Corsi and Fenwick are the best indicators of possession that exist, but I say they aren't good enough. It won't be long and we'll have real possession data once the players and puck are all chipped up, and my contention is that we'll find out that corsi and fenwick weren't really tracking much of anything useful.

Perfectly worded, and this has been one of my contentions with corsi. At it's worst it shows strong possession stats for a team that cycles the puck decently and then gets a lot of floater/weak shots on net once the cycle runs out, and at it's best it shows that winning teams have the puck a lot....excellent, yes they do.

[DJones]

I know this has been beaten to death but if someone could answer these questions for me that would be great. Maybe I'm looking at the whole thing wrong.

Ok so say with changing nothing else, we magically got Price to play for the Flames last year. It's a fair assumption that he could have raised our save percentage and therefore our PTO by 1. Him saving an extra shot out of a hundred for the Flames should have no effect on our shooting percentage. So we would bump our PTO by one and there should be no reason to expect regression anymore than there would be if Hiller was in net. I don't get why it's considered luck, it should be wow that team has a high PTO they must be doing something right.

In basketball when teams have high percentages it is considered skill, but in hockey the assumption is that everyone should fall to the mean eventually. The whole idea of goals being random events is flawed. They're trying so hard to find the "on base" equivalent of baseball. Problem is that in baseball getting on base is a tangible event that carries through to the next play consistently and effectively. If you take a shot in hockey it has no concrete barring on any future event. This is just a more complicated case of people making the incorrect assumption that if there was 4 reds in a row black must be overdue. Every shot does not have 91% chance to go in. It has anywhere from 1% to 99%. Treating each shot as equal is a waste of brain power.

[GranteedEV]

Originally Posted by DJones

In basketball when teams have high percentages it is considered skill,

Even with our high PDO (100.9 or 6th best in the league including playoffs) at 5 vs 5, the Flames scored 49% of the goals(20th best in the league). So on average, we were out-goal-scored at 5 vs 5. Even with their average PDO (100.2 or 15th best), the Kings out-goal-scored teams at 5 vs 5 (55.0%, or 4th best in the league).

Playoff hockey is mostly 5 vs 5, there's no 4 on 4 overtime and as we learned the hard way you don't quite get the power plays as favorably as you hope. If they were not terrible at 4 vs 4 (20.0% Goals for percnetage) and shootouts, the Kings would have made the playoffs and been a cup favorite.

If you want to compare that to basketball, it's like getting hammered on rebounds but making your threes. Eventually the team that got 20 more shots tends to have an edge even in basketball. It's no different... efficiency has its place but so does opportunity. The Flames did not give themselves much opportunity to outscore the other team unless it was special scenarios like drawing powerplays, not taking penalties, and pulling Hiller/Ramo.

[Street Pharmacist]

Originally Posted by DJones

I know this has been beaten to death but if someone could answer these questions for me that would be great. Maybe I'm looking at the whole thing wrong.

Ok so say with changing nothing else, we magically got Price to play for the Flames last year. It's a fair assumption that he could have raised our save percentage and therefore our PTO by 1. Him saving an extra shot out of a hundred for the Flames should have no effect on our shooting percentage. So we would bump our PTO by one and there should be no reason to expect regression anymore than there would be if Hiller was in net. I don't get why it's considered luck, it should be wow that team has a high PTO they must be doing something right.

In basketball when teams have high percentages it is considered skill, but in hockey the assumption is that everyone should fall to the mean eventually. The whole idea of goals being random events is flawed. They're trying so hard to find the "on base" equivalent of baseball. Problem is that in baseball getting on base is a tangible event that carries through to the next play consistently and effectively. If you take a shot in hockey it has no concrete barring on any future event. This is just a more complicated case of people making the incorrect assumption that if there was 4 reds in a row black must be overdue. Every shot does not have 91% chance to go in. It has anywhere from 1% to 99%. Treating each shot as equal is a waste of brain power.

The part is think your missing is one that is at the heart of the whole debate:

How much is skill, and how much luck?

As you've mentioned, PDO is simply shooting % plus save %. Intuitively they should measure skill. However, hockey is a highly random game. A goal comes once in every 8-13 shots on average. Even the greatest of snipers can't score at will or get the same quality chances every time. That then means they're going to be times where a lot more than the players average go in and times where a lot less than average go in.

If you're then looking at a single data set, say one season, there's going to be teams scoring way more than average, and some less. High and low PDO are rarely repeated. That would likely mean then that most high PDO are due to being in a data set that is higher than the team's real average.

[DJones]

Originally Posted by GranteedEV

The best indicator of possession that exists is measuring possession time. With the chipped pucks the NHL is introducing that's probably going to be phased in. Regardless, it doesn't matter if Fenwick etc are not perfect. They're highly correlated with winning:

5vs5 Fenwick Percentage, NHL Stanley Cup Champions:

2015 Hawks - 52.7 (7th best)
2014 Kings - 56.1 (Best)
2013 Hawks - 54.9 (3rd best)
2012 Kings - 53.6 (4th best)
2011 Bruins - 50.1 (16th best)
2010 Hawks - 57.8 (Best)
2009 Penguins Post-Thierren Firing - 53.4 (6th best)
2008 Red Wings - 59.0 (Best)

As far back as the stats exist the only team to win the cup that wasn't very good in the Fenwick stat was a Bruins team that most feel only won because Tim Thomas stole it for them in what might be the best goaltending performance ever.

5vs5 Fenwick Percentage, 2015 Flames - 45.8 (28th best)

I'd love to believe we can defy logic, but we're not yet even close to that Boston team that got by on goaltending. Does it really matter if it's a 1:1 ratio of corsi to possession? The real takeaway is that Cup champs are good at it, and right now the Flames are not. The Flames also were not cup champs to disprove that.

ONLY getting better at just Corsi won't make the Flames a better team, but we need to keep adding good hockey players and cutting bait with bad ones without getting emotionally attached. Brad T's already shown a willingness to do so with Frolik and Hamilton... he clearly didn't think Colborne/Jones or Wideman were "good enough" in their roles last year.

So from 1st to 16th have all been seen in 8 years. Your stats teacher would fail you in a heart beat if you ever said that was highly correlated. There is all the likelihood in the world that Corsi has something related to the actual reason for winning but it is no way the end all be all. So 1 year is getting thrown out for goal tending and then you are left with 2 teams winning 5 out of 7 years. Whose to say LA and Chicago aren't really good teams that happen to have a theory of shooting as much as possible. Other teams are mimicking that at the moment and are also having success. There is no evidence whatsoever that that is going to continue forever. Same logic stated that you needed a big man to win in basketball or you need a dominant defense to win in the NFL. People always find new ways to work the system.

And no one is saying that Calgary should have won the cup last year. I had them at about the 14th best team. I had Anaheim being the best team for most of the year and they weren't great and LA I had as being an average team and they were the best.

Calgary wasn't an elite team for a lot of reasons but their shot differential wasn't one of them. Could have easily taken more shots and not collapsed on our net as much. I don't think that would have won us anymore games.

[DJones]

Originally Posted by Street Pharmacist

The part is think your missing is one that is at the heart of the whole debate:

How much is skill, and how much luck?

As you've mentioned, PDO is simply shooting % plus save %. Intuitively they should measure skill. However, hockey is a highly random game. A goal comes once in every 8-13 shots on average. Even the greatest of snipers can't score at will or get the same quality chances every time. That then means they're going to be times where a lot more than the players average go in and times where a lot less than average go in.

If you're then looking at a single data set, say one season, there's going to be teams scoring way more than average, and some less. High and low PDO are rarely repeated. That would likely mean then that most high PDO are due to being in a data set that is higher than the team's real average.

Every sport is a highly random game but no other sport makes those types of nonsense assumptions.

Good defense messes up SH% in basketball. A good QB with a strong running game can destroy a great defense. Some guys just know how to hit fastballs. Sure it doesn't help everytime but no other sport calls it luck.

What if Anaheim would have won the cup this year? Haven't seen anyone say the wouldn't have been worthy. Not the greatest goalie. Just a good team with a 17th place Corsi. They're my favorite for the cup next year and they aren't an elite possession team. Clearly something else is going on.