All Purpose Analytics and Extended Stats Discussion

[Hivemind]

My concern with SportsVU is with the detail of analysis provided may make Oates' micromanagement look tame. Hockey may lose it's creativity and become robotic.

http://www.sportvu.com/PDFs/ESPN_62011.pdf

Has basketball or soccer become any more "robotic?"

Just because they can track these things won't instantly turn the players into machines. The players will still need to execute.

 

[g00n]

Has basketball or soccer become any more "robotic?"

Just because they can track these things won't instantly turn the players into machines. The players will still need to execute.

I can see them using such info to train and adapt. Just like teams adapt to rule changes. Someone will find a way around any attempt to overly stifle play.

 

[Hivemind]

Worth noting that, as best as I can recall, it hasn't been confirmed SportVU will be the actual company used. I believe it's only been noted that a SportVU-like technology (ie, camera based) will be installed and used.

 

[RandyHolt]

I like my PIMs and +- thank you very much. White bread and water. That said, I know little here. Heck, I am still pissed sorting by PIMs gives the folks with 0 at the top

Is there chance a player could focus on boosting his stats, with the intent on feeding the statisticians unbeknownst to them, knowing its going to be the gateway to a big paycheck?

Be a puck hog, don't dump it in, or, shoot it on goal at every chance.... what can a player easily do that will launch him to fancy stat stardom?

#FSS

 

[ChibiPooky]

I like my PIMs and +- thank you very much. White bread and water. That said, I know little here. Heck, I am still pissed sorting by PIMs gives the folks with 0 at the top

Is there chance a player could focus on boosting his stats, with the intent on feeding the statisticians unbeknownst to them, knowing its going to be the gateway to a big paycheck?

Be a puck hog, don't dump it in, or, shoot it on goal at every chance.... what can a player easily do that will launch him to fancy stat stardom?

#FSS

Tough to say. If he's making bad hockey plays just for stats, the coach should reduce his ice time and role, giving him less opportunity to make those plays.

 

[g00n]

Tough to say. If he's making bad hockey plays just for stats, the coach should reduce his ice time and role, giving him less opportunity to make those plays.

True, but what happens when they argue back that they're being judged based on these stats, so why shouldn't they be trying to boost them especially if said stats are supposedly indicative of positive possession and behavior?

Players in all sports are aware of this kind of thing. It will start happening the more the stat becomes understood and acknowledged. Since teams are hiring people to manage these stats you can bet the players are taking notice.

 

[Hivemind]

If the coach buys that argument, hopefully someone from the analytics department will come down and smack them all upside the head.

 

[fedfed]

As far as I know, the main reason why the stats revolution hasn't happened in hockey is because teams don't want it to be used at the arbitration hearings.

 

[NobodyBeatsTheWiz]

As far as I know, the main reason why the stats revolution hasn't happened in hockey is because teams don't want it to be used at the arbitration hearings.

The stats revolution hasn't happened in hockey because the stats are good enough yet, and I'm skeptical they ever will be.

 

[BobRouse]

I know there is some question in regards to the importance of face offs via fancy stats.

However early in the game last night we got dominated at the dot and it wasn't until we evened things out that possession went in our favor.

It just seems to stand to reason that if you win face offs you will have the puck more thus generate more shots thus generate more goals thus generate more wins.

 

[ChibiPooky]

I know there is some question in regards to the importance of face offs via fancy stats.

However early in the game last night we got dominated at the dot and it wasn't until we evened things out that possession went in our favor.

It just seems to stand to reason that if you win face offs you will have the puck more thus generate more shots thus generate more goals thus generate more wins.

Some people have tried to help cut down this sort of noise by eliminating events within [x] seconds of a stoppage in play. I don't like this as it eliminates a fair number of events from a pool that's already almost too small to be meaningful (since a lot of the value of these stats comes from the ability to say "over a large enough sample"). I think things are coming in the way of properly adjusting stats for faceoffs, but the focus is probably (rightly) on adjusting for quality of competition, which I think is a far bigger factor. I'd also like to see some efforts toward adjusting for role (i.e. checking line vs scoring line, stay-at-home D vs offensive guy, etc).

 

[BobRouse]

Some people have tried to help cut down this sort of noise by eliminating events within [x] seconds of a stoppage in play. I don't like this as it eliminates a fair number of events from a pool that's already almost too small to be meaningful (since a lot of the value of these stats comes from the ability to say "over a large enough sample"). I think things are coming in the way of properly adjusting stats for faceoffs, but the focus is probably (rightly) on adjusting for quality of competition, which I think is a far bigger factor.

OK Chibi...remember who you're talking to...I don't speak Chinese so have no idea what you just said

What is your take on faceoffs and their importance?

We have seemed to be a below average faceoff team since the departures of Gordon and Steckel and it does seems that is ONE reason for our declining possession.

Or am I way off here?

 

[ChibiPooky]

OK Chibi...remember who you're talking to...I don't speak Chinese so have no idea what you just said

What is your take on faceoffs and their importance?

We have seemed to be a below average faceoff team since the departures of Gordon and Steckel and it does seems that is ONE reason for our declining possession.

Or am I way off here?

I do think faceoffs are important. I don't know how important, but here's the structure of how I'd find out (since I have neither the data nor the time to compile it):

1) Reverse what those "remove the events within [x] seconds of a stoppage" guys are doing - instead, look at only those events
2) Find a multi-year leaguewide average for how the faceoff-winning team fares vs how the faceoff-losing team fares within those [x] seconds
3) Calculate the leaguewide average "faceoff" time per 60
4) Multiply, and that's more or less the impact you expect faceoffs to have on a team's overall possession
Caveats) Faceoff "wins" are scored interestingly and there may need to be adjustments based on this. There is definitely a boost that comes from having the puck to begin with (faceoff win impact trickling into "open play"), and I have no idea how to account for that or how significant it is.

 

[BobRouse]

I do think faceoffs are important. I don't know how important, but here's the structure of how I'd find out (since I have neither the data nor the time to compile it):

1) Reverse what those "remove the events within [x] seconds of a stoppage" guys are doing - instead, look at only those events
2) Find a multi-year leaguewide average for how the faceoff-winning team fares vs how the faceoff-losing team fares within those [x] seconds
3) Calculate the leaguewide average "faceoff" time per 60
4) Multiply, and that's more or less the impact you expect faceoffs to have on a team's overall possession
Caveats) Faceoff "wins" are scored interestingly and there may need to be adjustments based on this. There is definitely a boost that comes from having the puck to begin with (faceoff win impact trickling into "open play"), and I have no idea how to account for that or how significant it is.

Why not keep it simple and see if their is a correlation between good faceoff teams and good possession teams?

Basic I know but I'd have to think it would mean something.

 

[g00n]

I posted links in the past showing FO are among the more positively correlated stats vs winning. Can't remember exact breakdown.

Qoc and role are crucial but how you measure qoc matters. If we measure it with the same terms we are trying to define the results may not be pure enough. For example if qoc is based on corsi while contributing to a corsi related stat. A high corsi factor in qoc might not actually mean a more difficult assignment in terms of getting shots off.

 

[ChibiPooky]

Why not keep it simple and see if their is a correlation between good faceoff teams and good possession teams?

Basic I know but I'd have to think it would mean something.

My guess is you wouldn't see a very strong correlation. Far more of the game is spent in "open play" than immediately following a faceoff, and winning faceoffs isn't a skill that can be used in any situation except an actual faceoff. So I have a sense that if you try to look at entire games, the effect is pretty diluted, resulting in weak correlation.

 

[BobRouse]

Well if Goon is right and Faceoffs are positively correlated to winning then it must mean they are positively related to possession since possession is positively correlated to winning right?

Over the years how many times have we seen faceoffs directly leading to goals? We saw a few last night in fact.

I've seen several where Backstrom would win one directly to Ovie or Semin who would snipe one home.

My gut feeling is that faceoffs are highly underrated and more important than we are led to believe.

 

[ChibiPooky]

I posted links in the past showing FO are among the more positively correlated stats vs winning. Can't remember exact breakdown.

Qoc and role are crucial but how you measure qoc matters. If we measure it with the same terms we are trying to define the results may not be pure enough. For example if qoc is based on corsi while contributing to a corsi related stat. A high corsi factor in qoc might not actually mean a more difficult assignment in terms of getting shots off.

Wanted to mention that QoC at the team level is largely ignored at the moment. We know that schedules are not created equal, even more so with the 7 team vs 8 team divisions.

 

[ChibiPooky]

Well if Goon is right and Faceoffs are positively correlated to winning then it must mean they are positively related to possession since possession is positively correlated to winning right?

Over the years how many times have we seen faceoffs directly leading to goals? We saw a few last night in fact.

I've seen several where Backstrom would win one directly to Ovie or Semin who would snipe one home.

My gut feeling is that faceoffs are highly underrated and more important than we are led to believe.

Two things can be positively correlated with a weak correlation. When you're looking for correlation, you generally construct a scatter plot of the two things you want to correlate, one on the x axis and one on the y axis, and you look for a linear trend. If that trend line has a positive slope, it's a positive correlation, with higher (steeper) slope meaning higher positivity of correlation. Strength of correlation is tied to how tightly the plotted events are grouped to the trend line.

Strong correlation

Weak correlation

 

[BobRouse]

Chibi...you my friend are a total NERD!

What I gather is that there is more than meets the eye than saying Good face off team=Good possession team=Good team

 

[g00n]

Well if Goon is right and Faceoffs are positively correlated to winning then it must mean they are positively related to possession since possession is positively correlated to winning right?

Over the years how many times have we seen faceoffs directly leading to goals? We saw a few last night in fact.

I've seen several where Backstrom would win one directly to Ovie or Semin who would snipe one home.

My gut feeling is that faceoffs are highly underrated and more important than we are led to believe.

Not need necessarily. I cant search now maybe another can find my post on it. Topic was hits vs winning.

 

[ChibiPooky]

Chibi...you my friend are a total NERD!

What I gather is that there is more than meets the eye than saying Good face off team=Good possession team=Good team

If you know of a place that sells quality pocket protectors, let me know. Impossible to find good stuff these days

But yeah. To put it another way, what you said (good face-off team = good team) is measuring the positivity of the correlation. Good faceoff team = bad team would be a negative correlation. Strength of correlation is measuring how close that is to actual =.

 

[QuadrupleDeke]

I question the accuracy of the measurements, but the numbers are probably generally close:

http://www.powerscouthockey.com/blog/

According to that article, Ovechkin has by far the fastest acceleration of any NHLer they tested:

The amount he beat everyone else by is a bit insane/unbelievable. Also, I never thought of him as a top-speed guy, but according to that he's in the upper quarter of NHL stars.



EDIT: According to that chart, it would take Ovechkin almost 4 hours to get to his top speed. Crosby almost 5 hours. Perhaps they meant km/hr/sec and not km/hr^2 ?

 

[HunterSThompson]

Well if Goon is right and Faceoffs are positively correlated to winning then it must mean they are positively related to possession since possession is positively correlated to winning right?

Over the years how many times have we seen faceoffs directly leading to goals? We saw a few last night in fact.

I've seen several where Backstrom would win one directly to Ovie or Semin who would snipe one home.

My gut feeling is that faceoffs are highly underrated and more important than we are led to believe.

Two things correlated to the same thing does not necessarily mean that those two things are related. Celery stalks and Jelly are both positively correlated with Peanut Butter as shopping habits go. That does not mean that people buy Celery when they buy Jelly as well. Just one of the reasons that correlation =\= causation.

 

[BobRouse]

Two things correlated to the same thing does not necessarily mean that those two things are related. Celery stalks and Jelly are both positively correlated with Peanut Butter as shopping habits go. That does not mean that people buy Celery when they buy Jelly as well. Just one of the reasons that correlation =\= causation.

Are you seriously suggesting eating bread doesn't cause hair loss??

I can assure you 99% of bald guys eat bread.