Brian Burke is a Dinosaur

[Man Bear Pig]

I might be the biggest baseball stat man around and no, Boston never won with moneyball. Moneyball is using metrics to exploit overlooked statistics. The only reason Beane used this strategy was because he didn't have a choice. Boston attempted this strategy after Beane had success in Oakland and abandoned it after a few years and began going on spending sprees.

 

[kmad]

Hockey probably won't ever reach the point of baseball, however when the NHL begins to track puck touches, possession, passes completed, etc., we'll see a big revolution. That stuff will explain a lot of the game.

Puck touches for how long? Who made room on the ice so this player could touch the puck? How do we quantify quality of competition to measure how long this player should be touching the puck on average in this situation? Same for quality of linemates? How can we assign variable importance to duration of possession based on quality of linemates, quality of opposition, zone of ice, time remaining on the clock, position of other players, position of opponents? How will we factor quality of ice into passes completed, not only on an arena basis, but on a per-period and per-game basis? And on and on and...

I just don't see it working.

 

[QuadrupleDeke]

Puck touches for how long? Who made room on the ice so this player could touch the puck? How do we quantify quality of competition to measure how long this player should be touching the puck on average in this situation? Same for quality of linemates? How can we assign variable importance to duration of possession based on quality of linemates, quality of opposition, zone of ice, time remaining on the clock, position of other players, position of opponents? How will we factor quality of ice into passes completed, not only on an arena basis, but on a per-period and per-game basis? And on and on and...

I just don't see it working.

It's already working.

 

[Mathletic]

Puck touches for how long? Who made room on the ice so this player could touch the puck? How do we quantify quality of competition to measure how long this player should be touching the puck on average in this situation? Same for quality of linemates? How can we assign variable importance to duration of possession based on quality of linemates, quality of opposition, zone of ice, time remaining on the clock, position of other players, position of opponents? How will we factor quality of ice into passes completed, not only on an arena basis, but on a per-period and per-game basis? And on and on and...

I just don't see it working.

It's always the same thing, all you have to do to begin with is sit down, ask yourself the good questions. What is it that I'm looking for, how can I measure it objectively or with as much nuance as possible and get going. It's basic scientific protocol. Sure you may not want to do it, but it doesn't prove it can't be done.

Some of the questions you're asking have already been answered. Also, you have to be open to other ways of explaining the game. The goal of objective analysis isn't to put numbers on a subjective view of the game. The goal is to find what truly impacts the game, to what effect and so on.

 

[Wizeman*]

I think the second Marlins championship was on the cheap . I would argue that team was a money ball team just trying to be competitive.

By the way, the Florida Marlins NEVER lost a playoff series in their entire history.

 

[Captain Mittens*]

I might be the biggest baseball stat man around and no, Boston never won with moneyball. Moneyball is using metrics to exploit overlooked statistics.

David Ortiz, Kevin Millar, Mark Bellhorn, and Gape Kapler disagree with you

....after a few years and began going on spending sprees.

They had already won by the time they started buying free agents . Look for yourself. The Red Sox won their first world series on 10/27/04. Before they won, the most Epstein spent on a free agent was 3 Mil a year

https://docs.google.com/spreadsheet...DEzMVVET29UMDRGZTVGc2Vvbjk0Y0E&hl=en_US#gid=1

 

[Man Bear Pig]

David Ortiz, Kevin Millar, Mark Bellhorn, and Gape Kapler disagree with you



They had already won by the time they started buying free agents . Look for yourself. The Red Sox won their first world series on 10/27/04. Before they won, the most Epstein spent on a free agent was 3 Mil a year

https://docs.google.com/spreadsheet...DEzMVVET29UMDRGZTVGc2Vvbjk0Y0E&hl=en_US#gid=1

Every team had/has guys like them, it doesn't mean they used the moneyball strategy. Look at the rest of the lineup.

 

[nickschultzfan]

Winning puck battles wins hockey games. You need to watch tape to get a sense of that in a player.

 

[kmad]

It's always the same thing, all you have to do to begin with is sit down, ask yourself the good questions. What is it that I'm looking for, how can I measure it objectively or with as much nuance as possible and get going. It's basic scientific protocol. Sure you may not want to do it, but it doesn't prove it can't be done.

Some of the questions you're asking have already been answered. Also, you have to be open to other ways of explaining the game. The goal of objective analysis isn't to put numbers on a subjective view of the game. The goal is to find what truly impacts the game, to what effect and so on.

But the only reason we can give any value to statistics whatsoever is by cross-referencing them with the eye test. In theory, takeaways are an indicator of defensive play. If I tell you Patrick Kane has more takeaways than Jay McClement, you don't think "Well Kane must be better defensively." You think, "Well, we need to augment the statistic to better indicate reality." That's the eye test telling you McClement is better defensively than Kane, and that's nothing more than an effort to change statistics to reflect what we already know to be the effective method - the eye test. To figure out what truly impacts the game, you need to constantly alter your statistics with adjustments to compensate for what you know to be true, because of the eye test.

 

[kmad]

We'll never have a statistic that will tell you that people are afraid to chase the puck into the corner with Scott Stevens or Chris Pronger. Hockey has too many elements that are intangible.

 

[GuineaPig]

We'll never have a statistic that will tell you that people are afraid to chase the puck into the corner with Scott Stevens or Chris Pronger. Hockey has too many elements that are intangible.

If a player has a trait that is truly "intangible", it doesn't make a difference on the game. What you described for Stevens and Pronger would show up in their stats, because their teams would have better puck possession than their peers when they were on the ice.

 

[Mathletic]

But the only reason we can give any value to statistics whatsoever is by cross-referencing them with the eye test. In theory, takeaways are an indicator of defensive play. If I tell you Patrick Kane has more takeaways than Jay McClement, you don't think "Well Kane must be better defensively." You think, "Well, we need to augment the statistic to better indicate reality." That's the eye test telling you McClement is better defensively than Kane, and that's nothing more than an effort to change statistics to reflect what we already know to be the effective method - the eye test. To figure out what truly impacts the game, you need to constantly alter your statistics with adjustments to compensate for what you know to be true, because of the eye test.

With your example of Patrick Kane has more takeaways than McClement you're implying that takeaways would be the only defensive play one could make. Which is not true. It can easily be tested through regression analysis by asking the question how are takeaways correlated with goals against? Or are they at all correlated? What we are trying to do here is identifying the plays/strategies that lead to fewer goals being scored against your team in a given situation.

I'm not sure I understand your point but the eye test is required in any scientific endeavour in a sense. To classify a play as a homerun you have to know what a homerun is. From that point on you can add all the nuances necessary. What are the dimensions of the stadium, wind, humidity and whatnot.

Then you can ask the question what strategies lead to most runs being scored: good batting avg, good slugging pct, plate patience and whatnot. The same can be done for hockey.

If you want to analyze the performance of a goalie you have understand in what position he was put. How much time was spent on the PP, PK, ES. From what area of the ice shots were taken from and so on. All those things can be analyzed objectively.

 

[kmad]

With your example of Patrick Kane has more takeaways than McClement you're implying that takeaways would be the only defensive play one could make. Which is not true. It can easily be tested through regression analysis by asking the question how are takeaways correlated with goals against? Or are they at all correlated? What we are trying to do here is identifying the plays/strategies that lead to fewer goals being scored against your team in a given situation.

But then it's your better judgment telling you that since we know McClement to be better defensively, we need to adjust our definition of a defensive player to include more of what McClement does.

Ultimately, the most important aspect of defensive play is positioning, and positioning is fluid and relies in large part on anticipation, which is wholly intangible. How can you assign a metric to gap control? How can you measure a player's ability to effectively neutralize a passing lane on a penalty kill?

If you want to analyze the performance of a goalie you have understand in what position he was put. How much time was spent on the PP, PK, ES. From what area of the ice shots were taken from and so on. All those things can be analyzed objectively.

Still not perfect. You need to measure velocity of shot, position of defenders, length of possession of puck leading to scoring chance, etc. Too many variables.

There are so many brilliant minds working on hockey analytics that I would have to think a set of reliable metrics would exist by now if they were possible.

 

[Mathletic]

But then it's your better judgment telling you that since we know McClement to be better defensively, we need to adjust our definition of a defensive player to include more of what McClement does.

Ultimately, the most important aspect of defensive play is positioning, and positioning is fluid and relies in large part on anticipation, which is wholly intangible. How can you assign a metric to gap control? How can you measure a player's ability to effectively neutralize a passing lane on a penalty kill?

It's not a question of judgement. What we're trying to explain is what leads to fewer goals being scored against your team in a given situation. What are some of the factors we know to affect that? The play of your goalie, number or rate of shots against your team, quality of shots, the situation your team is on (i.e. discipline), and so on. We can now track where players are on the ice, where shots taken from, what distance there is between players etc. objectively. From that point on, through statistical methods you can quantify how important each action is.

Like I said, most the things you're talking about are measurable.

Still not perfect. You need to measure velocity of shot, position of defenders, length of possession of puck leading to scoring chance, etc. Too many variables.

There are so many brilliant minds working on hockey analytics that I would have to think a set of reliable metrics would exist by now if they were possible.

I never mentionned the word perfection. The question to ask ourselves is not whether or not analytics will lead to perfect decisions. The question to ask ourselves is whether or not we can make better decisions with analytics.

Btw, I hope you realize most "intangiles" you're talking about are quite tangible. Velocity is measurable, position on the ice is verifiable, time of possession (time) is measurable and so on. I'm not even getting into whether or not that's the direction a team would want to get into but that's another story.

Whether you like it or not there is some good work that has been done on hockey. That said, the ressources invested in hockey are nothing compared to what's being invested in basketball and baseball. I agree with you that there's tons of work to be done. Hence, that's why I think there's great opportunities for open-minded teams looking for an edge in building a winning team.

 

[Captain Mittens*]

Every team had/has guys like them, it doesn't mean they used the moneyball strategy. Look at the rest of the lineup.

The bolded part of your reply tells me that you don't know what you are talking about

look at how they were acquired and the price paid to acquire them.
That is what Moneyball is about. Finding undervalued players and maximizing the return on investment

 

[ldnk]

He's not wrong on the stats thing. A team that is built entirely on statistics is going to be a pretty mediocre team. It's the John Hollinger (NBA creator of the PER) approach to the game. It ignores character, team chemistry and takes a blanket approach to production.

It's also foolish to base your team on "intangibles" because it means you stick a guy like Dion Phaneuf as your captain on the basis of changing an IPod rather than how he commands the locker room at hockey-related events.

As for the Red Sox and Moneyball. Sure, they used sabremetrics in building their team, but the concept of moneyball was to make the best team you could with your salary. The Red Sox were not spending money efficiently. They bought a good team and augmented it with money-ball philosophies. It's a big difference. When Billy Beane started using the moneyball approach he had a team that was build almost entirely on the money-ball concept. They didn't have half a roster of very good, bordering on all-star or hall of fame calibre players with large contracts. They had guys who they acquired entirely because of their OBP.

 

[The Mighty Duck Man]

He's not wrong on the stats thing. A team that is built entirely on statistics is going to be a pretty mediocre team. It's the John Hollinger (NBA creator of the PER) approach to the game. It ignores character, team chemistry and takes a blanket approach to production.

It's also foolish to base your team on "intangibles" because it means you stick a guy like Dion Phaneuf as your captain on the basis of changing an IPod rather than how he commands the locker room at hockey-related events.

As for the Red Sox and Moneyball. Sure, they used sabremetrics in building their team, but the concept of moneyball was to make the best team you could with your salary. The Red Sox were not spending money efficiently. They bought a good team and augmented it with money-ball philosophies. It's a big difference. When Billy Beane started using the moneyball approach he had a team that was build almost entirely on the money-ball concept. They didn't have half a roster of very good, bordering on all-star or hall of fame calibre players with large contracts. They had guys who they acquired entirely because of their OBP.

It's funny that you mention Hollinger, especially in a thread that has championed the Dallas Mavericks as a "moneyball team". Hollinger hated the Mavs, especially in that championship season, and his power rankings always ranked them lower than their place in the standings. Cuban uses all sorts of approaches, but let's not act like they were basketball's Oakland Athletics. They stressed defense, hardly an advanced concept.

Also, regarding the Red Sox, they didn't exactly win because of sabermetrics either. They won largely because of their payroll, as others have said. Yes, they did integrate sabermetrics and it largely helped, but the biggest reasons were their high profile stars. Now, people like to mention Ortiz, as Bill James was the guy who wanted him. Don't get me wrong, I can easily see why James liked him at the time, but if the Sox got the Ortiz Bill James thought they were getting, they wouldn't have won jack.


I do like looking at advanced stats, it helps a little, but they really won't play a huge factor in hockey. Scouting can tell you many things you'd learn from them, so while it's nice for us fans, I don't think it plays a huge part for teams. The biggest thing I take from them is the ability to see if someone's normal stats are inflated, due to things like quality of competition or zone starts. But if you're watching that player every game, you probably see that already.

Not to say they should be ignored, but I'm really not a fan of CORSI, which consists of very little context, and that seems to be the main advanced stat in hockey. Also doesn't help when you have a guy like Jay Feaster, who seems to be a pretty big fan of advanced stats, is also terrible at building a team and was very close to making one of the worst moves in NHL history yesterday. Overall, there really isn't much of an example of advanced stats working out much in the NHL, so I can definitely see why some are hard to get behind them.

 

[Hero]

Hockey is not a sport where you can isolate stats lke you can in baseball. It's rarely 1 vs 1 settings in hockey, not often enough to compile reliable stats on them. Most of them game is played in multiple player vs multiple player settings.

 

[Mathletic]

It's funny that you mention Hollinger, especially in a thread that has championed the Dallas Mavericks as a "moneyball team". Hollinger hated the Mavs, especially in that championship season, and his power rankings always ranked them lower than their place in the standings. Cuban uses all sorts of approaches, but let's not act like they were basketball's Oakland Athletics. They stressed defense, hardly an advanced concept.

Also, regarding the Red Sox, they didn't exactly win because of Sabremetrics either. They won largely because of their payroll, as others have said. Yes, they did integrate Sabremetrics and it largely helped, but the biggest reasons were their high profile stars. Now, people like to mention Ortiz, as Bill James was the guy who wanted him. Don't get me wrong, I can easily see why James liked him at the time, but if the Sox got the Ortiz Bill James thought they were getting, they wouldn't have won jack.


I do like looking at advanced stats, it helps a little, but they really won't play a huge factor in hockey. Scouting can tell you many things you'd learn from them, so while it's nice for us fans, I don't think it plays a huge part for teams. The biggest thing I take from them is the ability to see if someone's normal stats are inflated, due to things like quality of competition or zone starts. But if you're watching that player every game, you probably see that already.

Not to say they should be ignored, but I'm really not a fan of CORSI, which consists of very little context, and that seems to be the main advanced stat in hockey. Also doesn't help when you have a guy like Jay Feaster, who seems to be a pretty big fan of advanced stats, is also terrible at building a team and was very close to making one of the worst moves in NHL history yesterday. Overall, there really isn't much of an example of advanced stats working out much in the NHL, so I can definitely see why some are hard to get behind them.

Cuban bought an adjusted +/- model developped Wayne Winston, which is used in many different ways, from skills adjusted +/- and other stuff. I agree he uses a ton of other ideas, that said, I think we should get passed the "moneyball" idea. The goal is objective analysis and getting some different data to make the best decisions possible.

Also, it's not like everybody in the analytics world agrees on everything. I'm not a big fan of Hollinger and don't think the models he has presented are that good.

As for hockey, you have to take those stats into context. Corsi is not a good measure for who is better than who. It's basically a differential of shots for and shots against. A proxy for time of possession.

I much prefer the adjusted +/- model developped by Brian MacDonald. I think it explains winning at an individual level much better.

 

[LeafDangler]

Regardless of where you stand on the utility of advanced metrics in hockey its the willfully ignorant close mindedness that bothers me. A GM should be always looking for an edge and be testing his philosophies against new ideas. Not resting on his laurels.

Anybody discarding traditional scouting solely for statistical analysis is a fool but the same goes vice versa. Both approaches offer information. I want a GM doing his damnedest to gather as much information as possible.

Furthermore can we refrain from talking about "Moneyball"? The majority of people in this thread seem to have a really difficult time grasping the philosophies it puts forward.

 

[Sybil227]

LOL, I wonder what Theo Epstein thinks of that middle quote.

...or Dean Lombardi, who brings up Moneyball in nearly every meeting with STH's.

 

[GIN ANTONIC]

Hopping in this thread halfway through but I'd say the Carolina Hurricanes were a good example of Moneyball. Picked up a lot write-off players from the scrap heap, signed them to good contracts and won it all. Sure there were other factors that went into but Justin Williams, Ray Whitney, Cory Stillman, Mike Commodore, Martin Gerber to name a few played big roles in raising the cup in 06. In fact I'm gonna make a thread about his on the main board.

Cheers

 

[It Was 4 to 1*]

 

[BoHorvatFan]

You clearly have no idea about the advanced statistics that went into the Dallas Mavericks winning their championship, especially the major tactical changes they made during the Finals to beat the Heat.

And the Red Sox hired Bill James, the author of the Baseball Prospectus that became the basis for all of moneyball, and then won 2 championships in 4 years.

So yes, those are my examples.

All that is true BUT there is nothing moneyball about having Manny Ramirez, a superstar hitter that signed the biggest contract ever at the time on your roster. Trading for Schilling and Beckett wasn't moneyball it was trading for ace pitchers that made a lot of money, signing Damon and Foulke weren't cheap bargains. I understand the high OBP, take a lot of pitches approach worked really well but without the star players they don't win.


To me moneyball is all about finding bargains not having a payroll over 100 million and trying to get on base a lot. Anyone can throw Pedro or Schilling or a younger Beckett on the mound and look great, and having a juiced up Manny and Ortiz dominating in the middle of the line up sure helped to.

 

[Mathletic]

All that is true BUT there is nothing moneyball about having Manny Ramirez, a superstar hitter that signed the biggest contract ever at the time on your roster. Trading for Schilling and Beckett wasn't moneyball it was trading for ace pitchers that made a lot of money, signing Damon and Foulke weren't cheap bargains. I understand the high OBP, take a lot of pitches approach worked really well but without the star players they don't win.


To me moneyball is all about finding bargains not having a payroll over 100 million and trying to get on base a lot. Anyone can throw Pedro or Schilling or a younger Beckett on the mound and look great, and having a juiced up Manny and Ortiz dominating in the middle of the line up sure helped to.

Then again, you could throw hundreds of millions to Barry Zito a Cy Young award winner. Some teams did their homework, the advanced stats just didn't back up the the general concensus that Zito and his great ERA was a great pitcher. The Red Sox spent their money wisely nonetheless. Like it's been said plenty of times already. Moneyball isn't only about winning on a budget. Moneyball is mostly about objective analysis and looking in a different direction than everybody is looking at. Manny Ramirez at 20MM a year may just be a bargain at that price.