News Article: "Fun With Numbers" - Advanced Stats Talk Here

[StefanW]

Yep. The more teams that do this, and the more analytics staff actually influence management, the more likely it is the Sens will fall behind. Melnyk's penny-pinching doesn't just impact the roster, but the front office too. Can't imagine Eugene is keen on spending money on more staff for analytics. Just look at the size of our scouting staff. Despite Melnyk constantly boasting about it, it's not that big. Of course size doesn't necessarily mean better results, but when everyone else has recognized the need for more information, and that people well-versed in that info are best for the work, it doesn't look good for the Sens.

I don't think it is about penny pinching at this point. The Leafs had money to burn for years without bothering to hire anyone in analytics. It really comes down to ideology, and whether the team believes they will get anything from it.

2 or 3 analytics guys on as advisors would be cheap. A full time department like the Leafs have is definitely the way to go, but it would probably be in the range of 500k per year to keep it running (guesstimate, no idea if I am right). If teams that use stats guys start to see more success, the ones that don't will be forced to hire to keep up.

 

[TheRightWay]

Well, if by possession numbers you mean Corsi%, its pretty close for career numbers. Karlsson 54.2% and Condra 53.9%.

Even in Karlsson's norris year it was close, with Karlsson at 54.9% and Condra at 54.1%.

This past year, Condra had 53.8% to Karlsson's 54.6%, so again not a huge gap.

It's pretty well established that Condra consistently puts up good to great possession numbers, so I'm not sure why you're calling him out on that. There are factors why Karlsson's possession numbers are more impressive, but that discussion is for another time.

No. By possession numbers I meant possession numbers. Corsi % is just one stat of many to take into account. The picture looks very different when accounting for zone starts and competition.

 

[FlyingJ]

I don't think it is about penny pinching at this point. The Leafs had money to burn for years without bothering to hire anyone in analytics. It really comes down to ideology, and whether the team believes they will get anything from it.

2 or 3 analytics guys on as advisors would be cheap. A full time department like the Leafs have is definitely the way to go, but it would probably be in the range of 500k per year to keep it running (guesstimate, no idea if I am right). If teams that use stats guys start to see more success, the ones that don't will be forced to hire to keep up.

True. I suppose the question we should ask is who on management right now is well versed in advanced stats, if anyone. Though I would still like to see 2 or 3 hires for this area specifically, just to give more info to Murray and co when making decisions.

 

[StefanW]

True. I suppose the question we should ask is who on management right now is well versed in advanced stats, if anyone. Though I would still like to see 2 or 3 hires for this area specifically, just to give more info to Murray and co when making decisions.

Yeah, that hits the nail on the head. When I see the Leafs building their current analytics department and I see the Sens doing nothing on this front I get a bit worried. The Leafs are our arch rivals, and if the stats guys have a say (not a given at this point) the team will not be making as many boneheaded moves like letting MacArthur walk. They usually have us on budget, but we always got by through savvy deals and better drafting. If they make up those differences and then some...well, lets just say I can handle off years better so long as the Leafs suck worse.

 

[Micklebot]

No. By possession numbers I meant possession numbers. Corsi % is just one stat of many to take into account. The picture looks very different when accounting for zone starts and competition.

I'll grant you that Karlsson tends to see tougher QOC, however he also has superior quality of teammates. As for zone starts, Karlsson is deployed in the OZ more frequently than Condra (makes sense given Condra is anything but an offensive threat).

So, no, the picture does not look very different. In fact, using zone adjusted stats (doesn't include the first 10 or so secs after a faceoff to negate the advantage/disadvantage of a OZ/DZ start), Condra actually ends up ahead of Karlsson in CF% for the career sample, and only moderately behind in last years numbers.

 

[FlyingJ]

Yeah, that hits the nail on the head. When I see the Leafs building their current analytics department and I see the Sens doing nothing on this front I get a bit worried. The Leafs are our arch rivals, and if the stats guys have a say (not a given at this point) the team will not be making as many boneheaded moves like letting MacArthur walk. They usually have us on budget, but we always got by through savvy deals and better drafting. If they make up those differences and then some...well, lets just say I can handle off years better so long as the Leafs suck worse.

6th Sens did a summary of a recent Randy Lee interview where he mentioned the Sens' thoughts on analytics. And while it eased some of my fears, it also left me feeling a bit disheartened:

"We do a lot of analytics-type stuff in different ways and we’ve been doing it for years. It’s not brand new, but it’s sort of taking on a new optic where teams are doing this and designating a certain person. I don’t think Bryan (Murray) believes in designating a certain person to this, but I think as an organization, we believe that if there are components of it that we think will help us and complement the job we do, we’ll try and incorporate it into our day-to-day work.”

IMO, I think having 2 or 3 people who are experts in this is important as reading them well can be hard. Just having familiarity with and using just some can lead to problems, like, as the post's author says, possibly ignoring some stuff with the belief that what matters more is they "know their players." That's when intangibles and liking someone for their personality or supposed locker room presence can get too much importance.

 

[Caeldan]

6th Sens did a summary of a recent Randy Lee interview where he mentioned the Sens' thoughts on analytics. And while it eased some of my fears, it also left me feeling a bit disheartened:

"We do a lot of analytics-type stuff in different ways and we’ve been doing it for years. It’s not brand new, but it’s sort of taking on a new optic where teams are doing this and designating a certain person. I don’t think Bryan (Murray) believes in designating a certain person to this, but I think as an organization, we believe that if there are components of it that we think will help us and complement the job we do, we’ll try and incorporate it into our day-to-day work.â€

IMO, I think having 2 or 3 people who are experts in this is important as reading them well can be hard. Just having familiarity with and using just some can lead to problems, like, as the post's author says, possibly ignoring some stuff with the belief that what matters more is they "know their players." That's when intangibles and liking someone for their personality or supposed locker room presence can get too much importance.

On the other hand, it could turn into the whole 'capologist' rage that we heard about a few years back. From what I can remember, it seems that the teams who visibly hired those are the ones that now are having serious cap issues.

 

[Micklebot]

6th Sens did a summary of a recent Randy Lee interview where he mentioned the Sens' thoughts on analytics. And while it eased some of my fears, it also left me feeling a bit disheartened:

"We do a lot of analytics-type stuff in different ways and we’ve been doing it for years. It’s not brand new, but it’s sort of taking on a new optic where teams are doing this and designating a certain person. I don’t think Bryan (Murray) believes in designating a certain person to this, but I think as an organization, we believe that if there are components of it that we think will help us and complement the job we do, we’ll try and incorporate it into our day-to-day work.â€

IMO, I think having 2 or 3 people who are experts in this is important as reading them well can be hard. Just having familiarity with and using just some can lead to problems, like, as the post's author says, possibly ignoring some stuff with the belief that what matters more is they "know their players." That's when intangibles and liking someone for their personality or supposed locker room presence can get too much importance.

A little bit of knowlege can be dangerous. Ideally, you want people who are both hockey guys and Anlalytics guys, not just specializing in one or the other. I think Toronto got that right with Dubas, though it's early to make that call.

 

[Micklebot]

On the other hand, it could turn into the whole 'capologist' rage that we heard about a few years back. From what I can remember, it seems that the teams who visibly hired those are the ones that now are having serious cap issues.

That could be a chicken or the egg situation: Are they in cap trouble because they hired a capologist, or did they forsee cap problems in the future, and try to address it proactively? They may have hired these guys because they knew that the team would be tight against the cap because of talen in the pipeline and the teams planned moves in the future.

Then again, if you're suggesting capologists are responsible for the nighmare contracts out there, that's another story.

 

[18Hossa]

A new stats article will be posted on the sens site on monday and will focus on Turris.

 

[StefanW]

A new stats article will be posted on the sens site on monday and will focus on Turris.

Can you please post a link to it in this thread when it appears?

 

[thinkwild]

I wonder how often an NHL mgmt/scouting team comes to a decision about a player, then looks at some advanced stats and changes their mind about the player. There are surely some interesting stats out there though. But they seem just as often used incorrectly as correctly so far as advanced hockey stats start coming out of their infancy.

I thought that was an interesting point about Corsi being more a team stat, or at least a line stat. But so sometimes is Sv%, when goalies have defenses great at boxing out leaving only perimeter shots or clearing rebounds so there are very few second shots. Or even face-offs, how come that is only applied to the centre. In fact, i bet if 20 of us sat at a game and recorded face-off wins and losses, no two of us would end up with them same results.

Data collection still seems proxy based, problematic, and inconsistent. I saw that clip recently about some company doing it for baseball, setting up a call centre like shop where people do 8 hour shifts recording the most minute aspects of every play of every game. Then they can sell that to teams. Great company to start for hockey, wouldnt want to work there though

Its a good point that Melnyk and his cost per point mantra should by definition be investing in a method for obtaining this data and be sure all his coaches and scouts are aware of it. But does he need to spend the money to compile that data? The video coaching budget i guess would go through the roof. Probably just as easy to outsource that.

Advanced stats so far seem interesting, the innovation is there, but the application seems like it's still being developed, but they are getting more and more interesting.

 

[Hammertyme]

A little bit of knowlege can be dangerous. Ideally, you want people who are both hockey guys and Anlalytics guys, not just specializing in one or the other. I think Toronto got that right with Dubas, though it's early to make that call.

I think your dead on. Hiring a bunch of pseudo knowledgeable bloggers who don't know the in and outs of the game (Yost) would be a total waste of effort and money. Most of the current stats are offensive/shot skewed. You need additional stats like passes made/missed(passer), completable passes gained/lost(receiver), battles won/lost, hits causing turnover, defensive deflections/blocked passes etc. You can rely on corsi but it is only the shots and a lot of things happen between shots. Possession is important but shots are only small part of the possession.
We need to see what happens between shots.

 

[StefanW]

Probably the most balanced article about the rivalry/feud between the analytics community and old school journalists. Hot off the presses, and already starting a firestorm on Twitter.
http://www.thehockeynews.com/blog/advanced-stats-vs-the-old-guard-inside-the-bitter-rivalry/

 

[BonkTastic]

Probably the most balanced article about the rivalry/feud between the analytics community and old school journalists. Hot off the presses, and already starting a firestorm on Twitter.
http://www.thehockeynews.com/blog/advanced-stats-vs-the-old-guard-inside-the-bitter-rivalry/

I feel bad for Simmonds, in a weird, sad sort of way.

The guy brings it on himself, but he just seems like such a bad representative of/ for old-school hockey mentality. There are far better and more well spoken members of the press who are still dubious of new statistics, yet still represent themselves like journalists and not spoiled teenagers having a temper-tantrum because their parent's don't agree that he should get a raise in his allowance. Who decided it would be Simmonds?... well, I guess Simmonds did. Shame. It's a completely unnecessary "war", with stubborn arses on both sides. The extreme on one side seems completely unwilling to let "non-hockey" people tell them they're wrong (at least, that's how it seems that they see it), and the other extreme seems so self-important and eager to turn the game on it's ear that they refuse to admit that anything that was done before they came along has any value.


That being said, Simmonds as the face of the anti-stats crowd comes across as like trying to have a reasonable debate about vaccines with a 16th century witch doctor.

 

[Nac Mac Feegle]

Problem is advanced stats have a hard time calculating skill and hockey IQ. You could have a bull of a player have possession of the puck for the full game, but if he has cement hands or doesn't have the IQ to know which net to shoot at, his advanced stats will probably look better than they should be.

You could also be a rope-a-dope team that simply skates up and down the boards and throws softballs at the net. Most teams won't worry too much about shots from horrible angles and will let you fire away all night.

Advanced stats are a great tool, but it's only one part of the overall picture.

 

[Micklebot]

Problem is advanced stats have a hard time calculating skill and hockey IQ. You could have a bull of a player have possession of the puck for the full game, but if he has cement hands or doesn't have the IQ to know which net to shoot at, his advanced stats will probably look better than they should be.

You could also be a rope-a-dope team that simply skates up and down the boards and throws softballs at the net. Most teams won't worry too much about shots from horrible angles and will let you fire away all night.

Advanced stats are a great tool, but it's only one part of the overall picture.

These kind of analogies are a little too extreme to be useful imo. I agree that Adv stats don't tell the whole story and likely never will, but this is like looking at a team and trying to judge them purely on how well they skate. Just like with old school evaluations, with adv stats you need to look at a lot more than just one criteria, so coming up with a criticism or situation where corsi, for example, gives an incomplete picture and using it to damn the whole method doesn't really add to the conversation, it just adds to the misunderstanding of how adv stats should be used.

All that said, because adv stats are in their infancy, the above is often who some are using them, which adds fuel to the fire.

 

[StefanW]

I feel bad for Simmonds, in a weird, sad sort of way.

The guy brings it on himself, but he just seems like such a bad representative of/ for old-school hockey mentality. There are far better and more well spoken members of the press who are still dubious of new statistics, yet still represent themselves like journalists and not spoiled teenagers having a temper-tantrum because their parent's don't agree that he should get a raise in his allowance. Who decided it would be Simmonds?... well, I guess Simmonds did. Shame. It's a completely unnecessary "war", with stubborn arses on both sides. The extreme on one side seems completely unwilling to let "non-hockey" people tell them they're wrong (at least, that's how it seems that they see it), and the other extreme seems so self-important and eager to turn the game on it's ear that they refuse to admit that anything that was done before they came along has any value.


That being said, Simmonds as the face of the anti-stats crowd comes across as like trying to have a reasonable debate about vaccines with a 16th century witch doctor.

I feel bad for him too, mostly because he has become such a target. The trolling and pot shots cut both directions. Based on what I have seen, which granted is very far from everything, the attacks on people who question analytics get very personal very quickly. Simmons makes reference in the article to people making fun of his appearance on his Twitter account, and I see continual jokes about intelligence. While it is frustrating to have the same debate again and again, rational and intelligent people can disagree about all sorts of things, including the meaning and application of statistical analyses.

I definitely agree that Simmons may not be the best person to articulate anti-analytics positions. At least not on his own. Someone who is better with numbers, and who can articulate criticisms of analytics without misrepresenting them, would be a far better choice.

Has anyone else noted that the supposed avalanche of analytics hirings over this past summer, which has been dubbed the summer of the blogger or other things along this same line, has amounted to 4 teams out of 30? I realize summertime reporting in hockey is what it is, but holy hyperbole batman. The big story is that the hirings left big holes in the online analytics community because the bloggers who were picked up made such a huge mark and were central to ongoing discussions (the article does make reference to this in passing). If we had to be completely honest, it remains to be seen how much of a mark these individuals will even have on their respective teams. They were hired as analysts, not GMs, and they will only provide input into decisions rather than making those decisions. I can't help but wonder if the online analytics community will feel their absence more than the teams that hired them will feel their presence.

 

[StefanW]

I guess it is time to put myself in the line of fire.

I wrote this piece on age and goaltender regression after reading reactions from the analytics community to the Anderson signing. I've posted it on the main area in the "By the Numbers" section asking for some feedback, and I will post it here as well. The feedback is important to me because I plan to do a bit of work with goaltender stats this coming season.

http://69.195.124.204/~integtd8/?p=141

Be nice

 

[mcnorth]

The problem with advanced stats is that there's no advanced stat to show how important they are.

 

[mcnorth]

I guess it is time to put myself in the line of fire.

I wrote this piece on age and goaltender regression after reading reactions from the analytics community to the Anderson signing. I've posted it on the main area in the "By the Numbers" section asking for some feedback, and I will post it here as well. The feedback is important to me because I plan to do a bit of work with goaltender stats this coming season.

http://69.195.124.204/~integtd8/?p=141

Be nice

Good stuff.

Glad to hear you're going to be doing some digging on goalies this upcoming season - that's my area of interest as well. Are you going to try to tackle save pct. and it's relation to winning pct.?

 

[StefanW]

Good stuff.

Glad to hear you're going to be doing some digging on goalies this upcoming season - that's my area of interest as well. Are you going to try to tackle save pct. and it's relation to winning pct.?

I'm going to start with focusing more on standard deviations of the save percentages, which is not an entirely new idea. Putting in a goalie with a lower save percentage good be the correct call if the standard deviation is lower, because it would mean the goalie is more consistent. If you are overmatched and really need a win, putting in a goalie with a higher standard deviation may make sense because that is the type of goalie who could steal a game for you. I also plan to do other work with refining goalie stats that I wont get into yet because I'm not sure if it will work out.

 

[mcnorth]

I'm going to start with focusing more on standard deviations of the save percentages, which is not an entirely new idea. Putting in a goalie with a lower save percentage good be the correct call if the standard deviation is lower, because it would mean the goalie is more consistent. If you are overmatched and really need a win, putting in a goalie with a higher standard deviation may make sense because that is the type of goalie who could steal a game for you. I also plan to do other work with refining goalie stats that I wont get into yet because I'm not sure if it will work out.

I look forward to reading your stuff as The Contrarian Goaltender is one of my favourite blogs. His work on win threshold and finding value by comparing goaltenders vs. their teammates' numbers was really interesting.

 

[StefanW]

I look forward to reading your stuff as The Contrarian Goaltender is one of my favourite blogs. His work on win threshold and finding value by comparing goaltenders vs. their teammates' numbers was really interesting.

Yeah, that is a great site. The work you mention is a lot like what is done in baseball.

I'm hoping I come up with something new. For now, the only thing that is certain is this is the last you'll hear of it if nothing comes up after about 20 games worth of data next season.

 

[Busboy]

Yeah, that hits the nail on the head. When I see the Leafs building their current analytics department and I see the Sens doing nothing on this front I get a bit worried. The Leafs are our arch rivals, and if the stats guys have a say (not a given at this point) the team will not be making as many boneheaded moves like letting MacArthur walk. They usually have us on budget, but we always got by through savvy deals and better drafting. If they make up those differences and then some...well, lets just say I can handle off years better so long as the Leafs suck worse.

The analytics department is less important to us than our prospect development department.

We have less money to spend on and off the ice, that's the reality. We need to allocate our resources to wherever suits us best and in our case it's prospect development.

It would be great to have some form of analytics but it's rather meaningless of we don't have the funds to make competitive offers to UFAs.