Tangotiger Blog

Wednesday, February 20, 2013

Grading Curve, or How Odds Ratio Method could have saved the day

I'm glad that the students were smart enough to beat such a basic curving scheme.

It's weird to set the starting point as the most extreme point, and work from there.  You could have a real outlier that would kill everyone below him.  Or,have no outliers at all, in which case, everyone would do great.  The kids realized that by mandating the non-outlier scenario, everyone would do great.

The ideal way is to set the MEAN as the starting point.  If for example you wanted the mean to be 85%, then that's what you do.  And then adjusting around that really is best done with the Odds Ratio Method.

Let's say for example you had 10 kids, and this was their actual scores:

Read More

Tuesday, February 19, 2013

Pedro a Hall of Famer?

?It's obvious that Pedro has the accomplishments of someone you (the Straight Arrow readers) would put in the Hall of Fame.

However, after which year would you say he was only a borderline candidate, that had he retired after year X, that you'd be on the fence with his candidacy?

And after which year Y would you say that he finally did enough to warrant a place in your Hall of Fame?

I do NOT want to talk about the actual Hall of Fame, or what the Holy Writers think.  I want to know what YOU think.

Monday, February 18, 2013

Methinks thou doth protest too much

?A reporter is found to fabricate quotes, then spends the day explaining how he misquoted himself.

Non-uniform playing area dimensions

?I asked Bill James about this issue.  Here was our exchange:

I think you are a proponent of baseball having non-standard dimensions for its parks. All the other major sports however have taken the opposite view of standardizing everything (though in the NHL, Boston and Chicago had smaller rinks until their new arens were built 20+ years ago). In light of the NFL thinking of expanding the field of play (side to side, closer to what the CFL has), would you support the idea that each team can set those dimensions as they want, within a league-imposed min/max range? I'd even go with non-standard field goal posts too. And why don't the leagues see it as one of baseball's charms that the parks are unique and capitalize on that?

Asked by: tangotiger

Answered: 2/17/2013

That would be my thinking as well.    From the standpoint particularly of basketball, I wouldn't think of it as one of baseball's charms; I would simply argue that it is better.   It is better from everyone's standpoint.   If you make the court wider, for example, you favor a smaller team with more quickness, and put a premium on ball-handling skills.   If you make the court more narrow, it favors big, burly guys, puts a premium on passing, and minimizes the importance of dribbling. Allowing different teams to experiment with different sized courts allows the game to breathe, allows the game to search out the most satisfying combinations.  Mandating one size for all courts makes the game rigid, unable to adjust. 

Skill v Luck

?Michael and Phil tag team to come up with a pretty good statement as to what we mean about skill and luck:

The standings are determined 60% by which teams are luckier than their opponents, and 40% by which teams are more talented than their opponents.

Injury begets injury

?If I am reading Pizza Cutter's chart correctly, it shows that a pitcher who had an injury last year (T-1) has a 73% chance of an injury this year (year T).  But, of those with no injury in T-1, they had only a 5% chance of injury in year T. If I had to guess, I would have set those numbers to 60% and 20%, or 70% and 30%, something like that.

Saturday, February 16, 2013

Ashton Kutcher bankrolled a gambling syndicate?

?He says so, and someone else confirms it.

"So if you can find the anomaly where the assumption is they're going to get heavy on one side, when the line moves it can actually move favorably as it pertains to the potential outcome of the game. So you set a margin of error against the line. If they've set it inaccurately, too hard to one side, you gauge that and play the other side of the line. You can gain probably a 40 percent statistical advantage on them.

It's really complicated. You gotta know a lot of data points — how this team played on various surfaces, in different weather — a lot. But these guys could do it. We were clearing, like, $750,000 in four weeks of college football. It was pretty fun. Then they caught on. The hypothesis had been that the house would just assume that I was a dumb actor with a lot of money who liked football."

...

Many believe Kutcher is one of the characters in Konik's highly-acclaimed book that details the author's time working for Rick "Big Daddy" Matthews, a gambler described as, "the world's smartest sports bettor and the mastermind behind the Brain Trust, a shadowy group of gamblers known for their expertise in beating the Vegas line."

It's widely believed, but never publicly acknowledged, that "Big Daddy" Matthews is Walters, a reclusive Las Vegas businessman with a reported net worth in the hundreds of millions. "My book 'The Smart Money' is a non-fiction memoir that includes an explanatory note about honoring privacy," Konik wrote in an email to The Linemakers. "I have no further comment."

Celebrities like Bruce Willis, Floyd Mayweather, Phil Mickelson and poker pro Phil Ivey have all been rumored to have been a part of Walters' syndicate at one time or another. Willis is also rumored to be one of the characters in "Smart Money," as is current Linemakers' analyst Richie Baccellieri. Baccellieri was a sports book manager at Caesars Palace in the mid-to late 1990s and is rumored to be the character "Stevie The Pencil" in "Smart Money."  Baccellieri would not comment on his or Kutcher's inclusion in the book.

Catapult Smartball

?Some inside view:

The in-ball module sits snugly inside a pouch with the ball’s interior bladder. Two beacons — one with a range of 16 to 47 inches and another with a range of 3 to 16 feet — pulse five times per second, sending data to the receiver. The receiver typically is strapped into a vest worn and sits comfortably between the player’s shoulders. The data recorder worn by the players can tell whether the player has the ball and can produce accurate measurements of possessions, speed, and distance.

The NAB Cup will mark the first time the ball-tracking tech has been used in an official game, but a few teams, including Gold Coast, North Melbourne and Adelaide, have tested the system during preseason practice. If all goes well, Catapult co-founder and COO Igor van de Griendt said he foresees the SmartBall being used at a future Super Bowl.

“We see the ball tracking as having enormous potential for Australian football and rugby this year, but have our sights on soccer and (American) football in the near future,” he said. “We’re all pretty excited about where the technology is headed.”

Friday, February 15, 2013

Plus/Minus is a GREAT concept

This post on Wages of Wins says that Plus/Minus is bad.  And I'm telling  you it's a great stat, if adjusted.  That post says:

 Indeed, we’ve seen iteration after iteration of “Adjusted Plus Minus” to improve upon the fact that all prior versions are bad at explaining wins. It’s a self-fulfilling prophecy. If you bring complicated metrics to try and solve a problem, you will remain convinced the problem is complicated and thus be convinced the only thing to solve the problem with is a complicated metric.

While I like plus/minus in hockey, I love it in basketball.  There's not enough scoring in hockey to get good samples, which is why we look for anything, like shots for and against.  In basketball, there's scoring a-plenty.

Anyway, this is the actual fact: we are recording who was on the ice/court/field when a run/goal/point is scored.  We are also recording this when you are NOT on the field of play.  This is actually at the heart of WOWY (with or without you).  And, as we've seen, it's fantastic.  And others have used that same process for cathcer framing. We use it all the time with ERA: we compare the pitcher's ERA to the rest of his team or rest of the league.  We do it with park factors.  Patriot has also done it with an adjusted "Wins Above Team".  It works fantastically well.

So, the idea of plus/minus is as old as analysis itself: what happens when the player is involved and what happens when he is NOT involved.

Now, the issue occurs when going from "involved" to "responsible".  You could in fact look at TEAM BABIP for a team when Andrelton Simmons is, and is not, playing SS.  This is a form of plus/minus.  The issue is of course that Simmons may be on the field when alot of balls are hit, except he's not responsible for most of them.  We solve this problem by putting Simmons on the field for 50,000 balls in play.  Unfortunately, that's a long career's worth, and we can't wait that long.

So, that's what we have: we have a signal, and we have the noise.  We make our adjustments to reduce the noise so that that signal can come through.

In 1986, Mark Howe was an astounding +85 (as was Brad McCrimmon equally high).  The Flyers were +74, meaning that when Mark Howe was not on the ice, the rest of the team was a negative.  Now, we don't know if it was Howe and/or McCrimmon, or the rest of the defensemen were terrible, but what we do know was that it was an enormous gulf.  There's signal in there.  There's noise too.

Which is why we live for sample size.  Which is why the work I did with catchers is as powerful as it is.  Not only do I have the sample size, but I also have the benefit that catchers have been exposed to dozens and in some case over a hundred, different pitchers in their careers.

If you don't like plus/minus, fine, whatever.  But calling it bad means you've come to a conclusion.  And that conclusion won't hold.

Winter Olympics v NHL

There are undoubtedly two marquee events at the Winter Olympics: figure skating, and men's hockey.  For figure skaters, the Olympics is a tremendous marketing opportunity for them.  They can parlay that exposure into skating "tours".  Hockey however is much different.  They already have their own league.  The exposure helps when they can show the games live and in normal viewing times.  Socchi, Russia doesn't help them so much.

In addition, the IOC has broadcast and distribution rights.  Whatever "Olympic ideals" is supposed to exist in the utopian world has to come face-to-face with the "absolute power corrupts absolutely" with our real-world Bayesian prior.  The NHL, and NHLPA, want their cut of the action.  So, now it seems they're trying to figure out the value for each party:

In return for sending its players to the Sochi Olympics, the NHL is trying to acquire video, photograph and website rights for the games. The IIHF and the IOC retain those exclusive rights now.

The NHL began sending its players to the 1998 Games in Nagano, Japan, and continued through the 2010 Vancouver Games. Even though the NHL received great exposure by having its players take part in an Olympics in North America, disrupting the season does come with a cost.

The stopping of the season, the potential injury risk to players, and no tangible upside for the NHL are all factors that create doubt about whether the investment is good for the league.

?

Thursday, February 14, 2013

Run Values by Run Environment

Using my Markov, we end up with these values. x-axis is the OBP, and the y-axis is the run value relative to all PA.

HOF, WAR, and years on the ballot

Very nice ?job by Dan.

Wednesday, February 13, 2013

Finnish-born GM in the NHL

First GM from Europe.

Kekalainen, who spent eight seasons with the NHL's St. Louis Blues from 2002-10 where he was involved in all facets of hockey operations, takes over a Columbus team that has not made the playoffs since 2009 and finished last overall in 2012.

Prior to working for the Blues, Kekalainen spent eight years with the Ottawa Senators as their top European scout and later as director of player personnel. He also played three seasons in the NHL with the Boston Bruins and Senators.

His resume reads like a standard resume of a GM, except for his birthplace?.  Given the huge number of European players in the NHL, it's kind of weird that he's the first.  I don't know how many hundreds of people have been GMs in the history of the NHL, but the odds that all of the best qualified had to be Canadian or American seems to be a stretch.

Similarly, how long before a woman takes the leap too?  They're already part of the CHL.  Hockey seems to be different from the other sports, as being a former player seems to be a pre-requisite.  Which tells me that it's either a highly specialized job, or it's likely a very inefficient selection process.

Adjusting PITCHf/x for Tropicana

Jon continues his work on calibration and adjustment. 

Tuesday, February 12, 2013

Anatomy of a soccer goal

http://www.youtube.com/watch?feature=player_embedded&v=YJuHC7xXsGA?

Contract Flexion

A contract extension is when a player's current deal is extended in terms of years, and usually with a higher rate per years.  Felix went from a 2/40 deal to a 7/175 deal.

But, let me make a crazy suggestion: what if it went the other way?  Not an extension, but a ... flexion.  Verlander also has a 2/40 deal.  What would it take for both sides to agree to rip up that deal, and sign a one year deal (MLBPA notwithstanding).  That is, if they agreed to a 1/10 deal, and then Verlander is a free agent, would both sides go for it?  How about a 1/5 deal?  1/1?

At what point does it make sense??

Get to know… Linear Weights

Steve does a good job by going through some various issues when using Linear Weights.?

Crowdsource v Actual Free Agents

Carson gives us an update.  ?

Mike Fast speaks

It's blocked at the office, but if someone can post a few interesting tidbits, that'd be cool.?

Monday, February 11, 2013

Boards: Custom Linear Weights