I agree and maybe it just my perception of the situation because I think that's where the problem and danger lies in the numbers boom. The people that are leading the charge understand the data and how to interpret the data but the don't understand the subject the data is defining. I see a lot of really bright and analytical minds trying to use data to define a game they don't completely understand so when they hit a point that the math just doesn't jive they either just create new math to explain why the original math didn't jive or they want to change the game so that it falls more in line with what their math wants. We have great math minds and we have great hockey minds but what we don't have is mind that understand both the math and the hockey. My fear is that as thing progress the game is going to be forced to change to meet the math as opposed to the math changing to meet the game which could fundamentally change hockey into a game that is very different than what it has always been.
- jaydogg1974

Its not a hockey issue. Its present in finance and academia and quite well understood. An academic might run 10,000 regressions on a data set to find something. And its really just overfitting.

I think its mammograms or something like that. But theres a medical test for finding cancer. If you have the cancer it will come back positive 90% of the time. If you don't have the cancer it will be positive 5% of the time. But maybe 1000 people don't have the cancer than those who do. So you get far more false positives than true positives. Most of the analytical guys don't under baysian probability or ignore it. Hockey probably has 3rd or 4th rate analytical guys (even in sports the guys in the NFL, NBA, and MLB get the first rate guys...and the really top guys are in finance or maybe academia). Rentech won't trade off one signal unless its a really good one. They often don't even know why markets are trading off one signal. But if 9 or 10 of there signals go off they will trade. Baysian probability. Maybe ones a false positive, but if enough are aligning then theres probably something there.

Also the hockey analytical guys often ignore causation and correlation. And think something that is correlating is a cause.

so the Stanley Cup we won, with a roster that was constructed using the help of these analytics (read HELP, not entirely), doesn't prove their value? We moved on from "old school" players like Adams, Glass, Sill and replaced them with strong analytic players and won the cup. That's not changing the game, that's evolving your mindset to ice the best team possible.
- buckster2342

so the Stanley Cup we won, with a roster that was constructed using the help of these analytics (read HELP, not entirely), doesn't prove their value? We moved on from "old school" players like Adams, Glass, Sill and replaced them with strong analytic players and won the cup. That's not changing the game, that's evolving your mindset to ice the best team possible.
- buckster2342

That was low hanging fruit. And we also got lucky with some old school scouting/coaching of developing mid round picks into solid hockey players. Which the analytics then showed were solid hockey players.

so the Stanley Cup we won, with a roster that was constructed using the help of these analytics (read HELP, not entirely), doesn't prove their value? We moved on from "old school" players like Adams, Glass, Sill and replaced them with strong analytic players and won the cup. That's not changing the game, that's evolving your mindset to ice the best team possible.
- buckster2342

When did I ever say they had NO value? I actually clearly pointed out that they are great for constructing teams and evaluating/comparing players.

They're not changing the game, the new trend is fans and front offices trying to fit numbers into arbitrary equations to come up with a rating to express a graphical representation of one player compared to his peers. The league is not trying to change the rules of the game or the way that it is played.
- jmatchett383

They may not be trying to change the game but when you change have you view the game and different aspects/skill become looked at and valued differently, change is inevitable.

Its not a hockey issue. Its present in finance and academia and quite well understood. An academic might run 10,000 regressions on a data set to find something. And its really just overfitting.

I think its mammograms or something like that. But theres a medical test for finding cancer. If you have the cancer it will come back positive 90% of the time. If you don't have the cancer it will be positive 5% of the time. But maybe 1000 people don't have the cancer than those who do. So you get far more false positives than true positives. Most of the analytical guys don't under baysian probability or ignore it. Hockey probably has 3rd or 4th rate analytical guys (even in sports the guys in the NFL, NBA, and MLB get the first rate guys...and the really top guys are in finance or maybe academia). Rentech won't trade off one signal unless its a really good one. They often don't even know why markets are trading off one signal. But if 9 or 10 of there signals go off they will trade. Baysian probability. Maybe ones a false positive, but if enough are aligning then theres probably something there.

Also the hockey analytical guys often ignore causation and correlation. And think something that is correlating is a cause.
- sditulli

You are overthinking this a bit. Corsi shows possession. What you do with the data is up to you. The same is true for all data as it relates to hockey, same as any other sport. Having the puck is better than not having it is the assumption. Now, when you begin to infer possession equals more goals you can accuse someone of correlating possession with more scoring. But it would be a fairly meaningless argument because....tis better to possess than to give away. Capish? Interpreting sports data requires some decent reasoning skills. Not everybody has those or understands deductive reasoning.

so the Stanley Cup we won, with a roster that was constructed using the help of these analytics (read HELP, not entirely), doesn't prove their value? We moved on from "old school" players like Adams, Glass, Sill and replaced them with strong analytic players and won the cup. That's not changing the game, that's evolving your mindset to ice the best team possible.
- buckster2342

Or as Grinder likes to the call them - the first line.

Gotta love reading the uninformed takes about analytics

You are overthinking this a bit. Corsi shows possession. What you do with the data is up to you. The same is true for all data as it relates to hockey, same as any other sport. Having the puck is better than not having it is the assumption. Now, when you begin to infer possession equals more goals you can accuse someone of correlating possession with more scoring. But it would be a fairly meaningless argument because....tis better to possess than to give away. Capish? Interpreting sports data requires some decent reasoning skills. Not everybody has those or understands deductive reasoning.
- powerhouse

I think you just summed up my points and proved my points.

Also Corsi could be misused. I can think of teams that would have awful possession stats, but be very good. If your team was based on strong counterattacks you might give up a ton of possession, but rarely give up odd man breaks while your offense is based on a quick transition and high danger scoring.

If you relied too heavily on corsi you might neglect guys who are snipers, but less athletic. Might not take a guy who sacrifices some possession, but has a much better shot.

I think you just summed up my points and proved my points.

Also Corsi could be misused. I can think of teams that would have awful possession stats, but be very good. If your team was based on strong counterattacks you might give up a ton of possession, but rarely give up odd man breaks while your offense is based on a quick transition and high danger scoring.

If you relied too heavily on corsi you might neglect guys who are snipers, but less athletic. Might not take a guy who sacrifices some possession, but has a much better shot.
- sditulli

If a guy takes 100 shots at 8% and another guy takes 80 shots at 10% which one is better?

If a guy takes 100 shots at 8% and another guy takes 80 shots at 10% which one is better?
- Feds91Stammer

Not enough information, but I'd probably take the 100 shot guy at 8% because most likely the guy taking more shots is in the offensive zone more so he's giving up fewer shots against. But that is a really simplified model.

Not enough information, but I'd probably take the 100 shot guy at 8% because most likely the guy taking more shots is in the offensive zone more so he's giving up fewer shots against. But that is a really simplified model.
- sditulli

Hence why Corsi is highly valued...

That was low hanging fruit. And we also got lucky with some old school scouting/coaching of developing mid round picks into solid hockey players. Which the analytics then showed were solid hockey players.
- sditulli

Old school mindset would be to take the biggest, meanest, toughest guy available. I dont think Rust, Wilson, Kuhnackle fit that description. Sheary was an UDFA who was labeled "Too small" for the NHL, which is why all these old school GMs passed on him. We got lucky getting him, and I'm not going to pretend I know the process of how we decided to sign him.

I'm sure they have analytics as part of their draft prep and scouting.

Hence why Corsi is highly valued...
- Feds91Stammer

What if 100 shots guy shoots wild on 80 of his shots and they lead to odd man rushes by the opposition and they score 50% of the time?

Who would you prefer then?

I think you just summed up my points and proved my points.

Also Corsi could be misused. I can think of teams that would have awful possession stats, but be very good. If your team was based on strong counterattacks you might give up a ton of possession, but rarely give up odd man breaks while your offense is based on a quick transition and high danger scoring.

If you relied too heavily on corsi you might neglect guys who are snipers, but less athletic. Might not take a guy who sacrifices some possession, but has a much better shot.
- sditulli

Generally, snipers are guys that elect to shoot more than pass. So their shot volume would be higher, as would their corsi for events. Last season, Henrique, Eriksson, and Panarin were the only 30+ goal scorers with under 200 shots. Most snipers don't pick their shots, they just shoot more often and from everywhere.

I think you just summed up my points and proved my points.

Also Corsi could be misused. I can think of teams that would have awful possession stats, but be very good. If your team was based on strong counterattacks you might give up a ton of possession, but rarely give up odd man breaks while your offense is based on a quick transition and high danger scoring.

If you relied too heavily on corsi you might neglect guys who are snipers, but less athletic. Might not take a guy who sacrifices some possession, but has a much better shot.
- sditulli

I was agreeing with you. I was trying to put it in laymen's term.

Hence why Corsi is highly valued...
- Feds91Stammer

Again your doing stupid over-simplification which does nothing but prove your an amateur analytics guy.

Generally, snipers are guys that elect to shoot more than pass. So their shot volume would be higher, as would their corsi for events. Last season, Henrique, Eriksson, and Panarin were the only 30+ goal scorers with under 200 shots. Most snipers don't pick their shots, they just shoot more often and from everywhere.
- j.boyd919

I don't really disagree. Though not sure I count those guys as pure snipers. They take a lot of shots because they are superior athletes who create a lot of shots therefore shoot more. But in theory you could have a guy with a superior shot by itself. It was nothing more than a very simplified comment to try to identify situations of correlation/causation and potential problems with painting too broad of a brush with analytics without really understanding what they are telling you.

so the Stanley Cup we won, with a roster that was constructed using the help of these analytics (read HELP, not entirely), doesn't prove their value? We moved on from "old school" players like Adams, Glass, Sill and replaced them with strong analytic players and won the cup. That's not changing the game, that's evolving your mindset to ice the best team possible.
- buckster2342

Pretty sure I read that the Hagelin trade was more to do with Sully having already coached him & he 'fit' the speed system Sully wants - never once did I hear or read it was assisted by analytics but I could be wrong.

Daley for Scuds was obvious & again doubt 'analytics' had anything to do with it.

The WBS 5 as noted many times on Gunners blogs didn't have good numbers for the NHL.

I'll go back to my question yesterday about LA. Are they still the analytic darlings they were when they won the Cup? Everyone knows they depend heavily on analytics?

If a guy takes 100 shots at 8% and another guy takes 80 shots at 10% which one is better?
- Feds91Stammer

What is a shot?