What’s the Point of "Analytics"?

[kappa123]

n/a

Click to read more...

 

[bigd]

When games become more predictable fans interest will wane. Accurate predictability will be the downfall of sports. Why watch when you can accurately predict the score through analytics?

 

[goosh]

When games become more predictable fans interest will wane. Accurate predictability will be the downfall of sports. Why watch when you can accurately predict the score through analytics?

Part of what analytics have shown in hockey is that there's a huge aspect of randomness/luck in the sport. That's here to stay. And that's okay. If you like sports based on pure skill, go watch chess.

 

[Doctor No]

Part of what analytics have shown in hockey is that there's a huge aspect of randomness/luck in the sport. That's here to stay. And that's okay. If you like sports based on pure skill, go watch chess.

Precisely - the *first* thing one should learn from an analytic study of sport is that there is a significant luck component (and I don't mean luck as in "things that we have not yet learned to measure", but truly luck).

Anyone expecting this to eventually boil down to an Excel exercise will be disappointed. To paraphrase one of my favorite ESPN quotes: "This team looks good on paper, but games aren't played on paper - they're played inside television sets."

 

[Mayor Bee]

Part of what analytics have shown in hockey is that there's a huge aspect of randomness/luck in the sport. That's here to stay. And that's okay. If you like sports based on pure skill, go watch chess.

The same goes with any sport. BABIP in baseball has demonstrated this, with some ability to predict what pitchers are likely to continue performing at a high level or regress back to mediocrity. The guy who started that train (Voros McCracken) published his original theory and findings in 1999, although other attempts to reach the same conclusion in a similar manner had existed previously.

 

[Mathletic]

When games become more predictable fans interest will wane. Accurate predictability will be the downfall of sports. Why watch when you can accurately predict the score through analytics?

Football and basketball (even more so) are quite predictable games. I doubt hockey will suffer from being more predictable if that's the future analytics hold hockey the sport. That said, what we've seen in baseball is quite the opposite. Analytics have led to a much more even field. Knowledge has spread throughout the sport and this past season has been as tight as any in the past and the field has tightened up for the last decade. If anything, that's the most likely outcome for hockey. If the sport becomes a game of perfect information, pretty much only randomness will dictate who wins and who doesn't. That is, if all teams actually embrace analytics.

 

[tacogeoff]

Precisely - the *first* thing one should learn from an analytic study of sport is that there is a significant luck component (and I don't mean luck as in "things that we have not yet learned to measure", but truly luck).

Anyone expecting this to eventually boil down to an Excel exercise will be disappointed. To paraphrase one of my favorite ESPN quotes: "This team looks good on paper, but games aren't played on paper - they're played inside television sets."

I take a look at the numbers every once in a while after the game but in reality it is more geared towards management and organizations but as doctor no said there is so much luck and variables invovled in the game that it may be hard to see this take full flight.


98% of my friends and associates could care less about the advanced numbers (those that know about them) that do not pertain to fantasy hockey pools.

 

[sweesh]

I think hockey is more immune to analytics than other sports because of the literal object being used to score with.

The puck - this little frozen rubber disk - while HIGHLY controllable per player in practice scenarios or small scrimmages etc is highly UN-controllable per player in today's ridiculously fast 5-on-5 game where every touch lasts mere seconds and ice conditions vary per barn and per randomized wear on surfaces at various points of games causing higher propensities for bounces and rolls and other aspects that cause that highly un-controllable object to react in ways entirely unpredictable given stick-blade angles, angle of attack, speed of skater, agility of same, hands of stick-wielder etc etc etc etc.

Meanwhile in baseball, the object, the rawhide covered ball is highly controllable for MUCH greater portions of play. There are fewer opportunities for variance though they are still there given differences between turf and grass and differences in moisture content or lack thereof in same. But largely hitters are sophisticated enough to influence where they impact the ball and where they attempt to put the object into play, just as pitchers have large swaths of control over where the ball goes and its physics. The baseball is much more manipulatable and thus analytics has more to say in that sport - just as an example that contrasts from hockey's randomness.

Analytics would mean more in a fully 4-on-4 hockey league because the added time and space would make the object, the puck, more controllable and less subject to randomizing aspects and thus player attributes would come back into play and game outcomes would become more forecastable.

I think analytics would be MUCH more useful than even in baseball and certainly more so than in hockey in the sport of BASKETBALL. Talk about a HIGHLY HIGHLY SUPER controllable scoring object with FAR fewer randomizing aspects outside of player attributes - you're talking about a highly grippable, visible ball in the hands of players for longer on surfaces uniform throughout the entire season of play with predictable bounce physics and feel and approach. Basketball is an analysts dream!

Football maybe too - though in rain or snow all bets are off - actually, maybe betting on underdogs makes most sense AFTER checking the weather reports and seeing a high probability of precipitation for the upcoming game. A wet and/or cold/hot football can be a highly random scoring object.

This is an interesting subject - thanks OP for starting it. I think I'm going to go write about this on my blog now (by pretty much copying this entire thing).

 

[Jtown]

the moment analytical data is applied, the more meaningless it becomes.


Analytics allows for people to understand why things happen. Why did the kings win the stanley cup as oppose to the florida panthers. Now that you have that data, the moment you apply it the scenrio is instantly different.

 

[Doctor No]

the moment analytical data is applied, the more meaningless it becomes.


Analytics allows for people to understand why things happen. Why did the kings win the stanley cup as oppose to the florida panthers. Now that you have that data, the moment you apply it the scenrio is instantly different.

Can you explain a bit further? I can't tell if you're trying to make a "Schroedinger's Cat" argument, or a "why bother?" argument, or what.

 

[Mathletic]

I think hockey is more immune to analytics than other sports because of the literal object being used to score with.

The puck - this little frozen rubber disk - while HIGHLY controllable per player in practice scenarios or small scrimmages etc is highly UN-controllable per player in today's ridiculously fast 5-on-5 game where every touch lasts mere seconds and ice conditions vary per barn and per randomized wear on surfaces at various points of games causing higher propensities for bounces and rolls and other aspects that cause that highly un-controllable object to react in ways entirely unpredictable given stick-blade angles, angle of attack, speed of skater, agility of same, hands of stick-wielder etc etc etc etc.

Meanwhile in baseball, the object, the rawhide covered ball is highly controllable for MUCH greater portions of play. There are fewer opportunities for variance though they are still there given differences between turf and grass and differences in moisture content or lack thereof in same. But largely hitters are sophisticated enough to influence where they impact the ball and where they attempt to put the object into play, just as pitchers have large swaths of control over where the ball goes and its physics. The baseball is much more manipulatable and thus analytics has more to say in that sport - just as an example that contrasts from hockey's randomness.

Analytics would mean more in a fully 4-on-4 hockey league because the added time and space would make the object, the puck, more controllable and less subject to randomizing aspects and thus player attributes would come back into play and game outcomes would become more forecastable.

I think analytics would be MUCH more useful than even in baseball and certainly more so than in hockey in the sport of BASKETBALL. Talk about a HIGHLY HIGHLY SUPER controllable scoring object with FAR fewer randomizing aspects outside of player attributes - you're talking about a highly grippable, visible ball in the hands of players for longer on surfaces uniform throughout the entire season of play with predictable bounce physics and feel and approach. Basketball is an analysts dream!

Football maybe too - though in rain or snow all bets are off - actually, maybe betting on underdogs makes most sense AFTER checking the weather reports and seeing a high probability of precipitation for the upcoming game. A wet and/or cold/hot football can be a highly random scoring object.

This is an interesting subject - thanks OP for starting it. I think I'm going to go write about this on my blog now (by pretty much copying this entire thing).

The most random game by a fair margin is baseball, then comes hockey, followed by football and then basketball. The 2 sports in which analytics play a big role are baseball and basketball. The 2 sports at each end of the spectrum. So, I disagree with your claim. Btw, you can verify that through Vegas odds. The most lopsided matchups in baseball like Kershaw pitching at home to some bad team will have a probability of maybe 70% of winning. The most lopsided games in hockey are a shade under 90%, a bit more for football and in basketball you get close to 99%.

...

Like the OP's point here: The NHL - and indeed professional sports in general - is currently mired an overabundance of "statistics" with a shortage of "analytics". Upcoming player/puck tracking data will only aggravate this situation, in which the progress of theory greatly lags behind the progress of observation.

Bunch of times you hear that once the SportsVu data will be given to teams, "analytics" people won't be needed as "regular" hockey people will be able to figure it all out. I tend to think it will be quite the opposite.

 

[Tawnos]

Like the OP's point here: The NHL - and indeed professional sports in general - is currently mired an overabundance of "statistics" with a shortage of "analytics". Upcoming player/puck tracking data will only aggravate this situation, in which the progress of theory greatly lags behind the progress of observation.

Bunch of times you hear that once the SportsVu data will be given to teams, "analytics" people won't be needed as "regular" hockey people will be able to figure it all out. I tend to think it will be quite the opposite.

Yeah, me too. I'm a big critic of the limits of our current set of data, although it's mostly because I see flaws in it and truly want there to be fewer flaws.

But as someone said to me recently, the more and more the stats gets used by the people running the clubs, the more and more those clubs will be using the same stats. Which means that when everyone has the same information, building a successful hockey club is going to come down to intangibles, group dynamics, and judgment of talent. Which is essentially what building a successful hockey club has always been about.

 

[Mc5RingsAndABeer]

When games become more predictable fans interest will wane. Accurate predictability will be the downfall of sports. Why watch when you can accurately predict the score through analytics?

We will never reach that point. There are far too many variables to account for.

 

[GirardIsStupid]

Does it help people who don't know how to watch hockey?

 

[Doctor No]

Does it help people who don't know how to watch hockey?

Can you elaborate?

I'll give you a short answer to what I think you're asking. The best way to truly understand hockey is with a combination of analytics and watching. Taking away either results in a loss.

Many people will deride those who use analytics with "oh, you've got your head in a spreadsheet. Try actually watching the games instead." This accomplishes several things: (1) they don't have to actually learn what's in the spreadsheets, (2) they can feel better about themselves, (3) by falling back to the "watch the games" argument, they've reached a point where they can never lose an argument, and (4) the "watch the games" argument has a hidden corollary that "I watch games better than you do".

Regardless, there are actually no individuals out there performing hockey analytics (*) who don't do so because they truly love the sport and enjoy watching it. We all "watch the game", as they say.

(*) I'm willing to admit the possibility of an outlier or two.

 

[wgknestrick]

Does it help people who don't know how to watch hockey?

You mean the "Watch the game" crowd?

If there is any single group that stands to gain the most from analytics in how to watch the game, it's the "watch the game" crowd. It probably sounds obvious and logical to most everyone here, but your ability to quickly evaluate meaningful opinions on players is augmented greatly when you start to learn "what" actually helps a team win games in the long haul, and "how much".

It's like watching cars go past you on a race track and trying to judge who's fastest without looking at the lap times. You just can't see the big picture without all the surrounding data. Your mind alone can't record that much data and process it. You can't watch 10 players at the same time.

Analytics is just a tool to help record and analyze the meaningful data of hockey. It is great because it is quantitative/numerical, always evolving, and just not binary (like the human mind). We advance when we start to treat things like they are a science. I really see the same struggle between the analytics and "watch the game" crowd as I do between the evolution and creationism.

 

[sweesh]

The most random game by a fair margin is baseball, then comes hockey, followed by football and then basketball. The 2 sports in which analytics play a big role are baseball and basketball. The 2 sports at each end of the spectrum. So, I disagree with your claim. Btw, you can verify that through Vegas odds. The most lopsided matchups in baseball like Kershaw pitching at home to some bad team will have a probability of maybe 70% of winning. The most lopsided games in hockey are a shade under 90%, a bit more for football and in basketball you get close to 99%.

...

Like the OP's point here: The NHL - and indeed professional sports in general - is currently mired an overabundance of "statistics" with a shortage of "analytics". Upcoming player/puck tracking data will only aggravate this situation, in which the progress of theory greatly lags behind the progress of observation.

Bunch of times you hear that once the SportsVu data will be given to teams, "analytics" people won't be needed as "regular" hockey people will be able to figure it all out. I tend to think it will be quite the opposite.

It sounds like my observation re: basketball was accurate, though I'll grant you I may be mistaken re: analytic usefulness for baseball vs. hockey - I will have to defer as I have not seen evidence for that, but that is not to say it doesn't exist. Where do you go to find credible data? (not a challenge, genuinely curious-thanks)

I do think however, 4-on-4 hockey would lend itself more to analytics than the current 5-on-5 set up - what's your opinion on that?

 

[Jtown]

Can you explain a bit further? I can't tell if you're trying to make a "Schroedinger's Cat" argument, or a "why bother?" argument, or what.

Sorry, I'd probably do a better job of explaining myself in a conversation but I'll give it a go.

Analytics is an after the fact analysis. For instance, after the season when the analysis is done, you will see that the playoff teams had a 60 percent success rate of entering the zone on their forehand, as oppose to 20 percent success rate on their backhand.

The next year " lets force teams to enter the zone on their backhand where they were only 20 percent successfull". The opposing player makes on the fly adjustments and realize he has an easier time entering the zone on his backhand . After the season when analysis is done the playoff teams had a success rate of 20 percent entering the zone on their forehand and 60 percent success rate on their backhand.

Athletes are looking for the path of least resistence to their goal. They don't think i shoot a higher percentage from this side of the ice as oppose to the other side of the ice, they think " I can get my shot off here as oppose to there" and as soon as you change your game plan based off analytics so does the opponent.

 

[Mathletic]

It sounds like my observation re: basketball was accurate, though I'll grant you I may be mistaken re: analytic usefulness for baseball vs. hockey - I will have to defer as I have not seen evidence for that, but that is not to say it doesn't exist. Where do you go to find credible data? (not a challenge, genuinely curious-thanks)

I do think however, 4-on-4 hockey would lend itself more to analytics than the current 5-on-5 set up - what's your opinion on that?

I build my own models for sports betting but all you have to do is look at the odds for games. They're posted online. For basketball you get odds like -1000, meaning there's over a 90% chance the team will win its game or that you have to win over 90% of the time a game with these odds to be profitable. You can access the same odds for the other sports.

I place my bets on this website

http://www.pinnaclesports.com/

NBA for tonight: http://www.pinnaclesports.com/League/Basketball/NBA/1/Lines.aspx

For instance, they evaluate the Clippers' chances at about 89.75% chance to beat the Lakers tonight.

the site may not be available for you if you're in the US.

 

[mkwong268]

Sorry, I'd probably do a better job of explaining myself in a conversation but I'll give it a go.

Analytics is an after the fact analysis. For instance, after the season when the analysis is done, you will see that the playoff teams had a 60 percent success rate of entering the zone on their forehand, as oppose to 20 percent success rate on their backhand.

The next year " lets force teams to enter the zone on their backhand where they were only 20 percent successfull". The opposing player makes on the fly adjustments and realize he has an easier time entering the zone on his backhand . After the season when analysis is done the playoff teams had a success rate of 20 percent entering the zone on their forehand and 60 percent success rate on their backhand.

Athletes are looking for the path of least resistence to their goal. They don't think i shoot a higher percentage from this side of the ice as oppose to the other side of the ice, they think " I can get my shot off here as oppose to there" and as soon as you change your game plan based off analytics so does the opponent.

This is a game theory problem, but eventually the set of strategy and counter strategy should get figured out then it'll come down to execution and being able to read the play so you can execute the correct counter plays. At that point the game should boil down to skill of players and luck (ie: execution errors) and no team will have an advantage from analytics.

 

[Doctor No]

Sorry, I'd probably do a better job of explaining myself in a conversation but I'll give it a go.

Analytics is an after the fact analysis. For instance, after the season when the analysis is done, you will see that the playoff teams had a 60 percent success rate of entering the zone on their forehand, as oppose to 20 percent success rate on their backhand.

The next year " lets force teams to enter the zone on their backhand where they were only 20 percent successfull". The opposing player makes on the fly adjustments and realize he has an easier time entering the zone on his backhand . After the season when analysis is done the playoff teams had a success rate of 20 percent entering the zone on their forehand and 60 percent success rate on their backhand.

Athletes are looking for the path of least resistence to their goal. They don't think i shoot a higher percentage from this side of the ice as oppose to the other side of the ice, they think " I can get my shot off here as oppose to there" and as soon as you change your game plan based off analytics so does the opponent.

I think that I see what you're saying, and if so, I agree. I said something in another thread a few weeks back that you may agree with:

One thing that I haven't seen explicitly mentioned here is the difference between using Corsi as a descriptive measure (how good is this team?) and using Corsi as a prescriptive measure (how can I make this team better?).

Stated differently, there's a correlation between Corsi and winning, but it's likely that winning teams have good Corsi levels as a result (since they're outshooting their opponents).

If one were to take this and say that existing teams should shoot more, because it will make them a better team, that part could not follow. In other words, being good causes shot differentials, not the other way around - and this may be what Drew Doughty was trying to say when he called Corsi "crap" (he's got a gift for brevity). Players don't go out and try to have good Corsi rankings - they go out and try to win, and a result of that is good Corsi rankings.

That was in a discussion about Corsi specifically, but I agree in general terms as well.

 

[Jtown]

This is a game theory problem, but eventually the set of strategy and counter strategy should get figured out then it'll come down to execution and being able to read the play so you can execute the correct counter plays. At that point the game should boil down to skill of players and luck (ie: execution errors) and no team will have an advantage from analytics.

1. Only 1 strategy can be used by one player each time. A dman can't play the blue line to prevent a zone entry and also be in position for a dump and chase.

2. Each strategy has a counter to it , so no matter what strategy is being employed there is a way around it.

An example would be in basketball, close proximity to the offensive player prevents uncontested jumpshots, but leaves one vulnerable to penetration. Far proximity from an offensive player prevents penetration but allows open jump shots.

 

[Mathletic]

1. Only 1 strategy can be used by one player each time. A dman can't play the blue line to prevent a zone entry and also be in position for a dump and chase.

2. Each strategy has a counter to it , so no matter what strategy is being employed there is a way around it.

An example would be in basketball, close proximity to the offensive player prevents uncontested jumpshots, but leaves one vulnerable to penetration. Far proximity from an offensive player prevents penetration but allows open jump shots.

Although you can only play one strategy at a time, you can still mix your strategies. You can look at this paper on soccer. It's a game theory approach to penalty kicks. Basically says that you should kick x% of the time to left (your strong side) and % of time to the right (weak side). If you deviate from that strategy, it will only come at your loss. Same for the goalie who should gamble to the left a certain % of time.

Professionals Play Minimax

http://www.palacios-huerta.com/docs/professionals.pdf

 

[Jtown]

Although you can only play one strategy at a time, you can still mix your strategies. You can look at this paper on soccer. It's a game theory approach to penalty kicks. Basically says that you should kick x% of the time to left (your strong side) and % of time to the right (weak side). If you deviate from that strategy, it will only come at your loss. Same for the goalie.

that my friend is not really game theory. That has it's roots in biomechanics, and exercise science.

 

[Doctor No]

that my friend is not really game theory. That has it's roots in biomechanics, and exercise science.

What Mathletic is describing is still a branch of game theory / optimization. My doctoral dissertation was in the latter.