Analysis Moneyball

[Turbocat]

It's definitely happening - every year in the Prospectus CD stats guys talk about their latest effort to isolate the value of taggers, or teams scoring efficiency, or whatever. But they also stress that CD is a business and that as a result this analysis remains confidential so it retains value. I doubt journos get a look at it. Most amateur analytics in US and European sport also relies on freely available data that is a lot more sophisticated than what we get here.

It's coincidental to this article but its interesting that a number of the most prominent amateur analytics guys in North American hockey have been hired this offseason by NHL clubs. Just as in Moneyball (arguably more so) there's been a pretty overt war of words between traditionalists and stats guys in the NHL and the hiring of a bunch of bloggers is seen as vindication of the bloggers. But really, they already won - they've kept it quiet, but dominant teams of late like the Kings and Blackhawks pretty clearly spend big money on their own analytics.

That said (while Im rambling) I also reckon the debate in hockey shows why you have to be careful of getting carried away with stats in and of themselves. The debate in hockey boiled down to an argument over whether or not possessing the puck is important. The answer is pretty obviously yes - the team that is better able to hold onto the puck will shoot more and score more goals over the course of a season. In much the same way, on-base percentage is a pretty simple, logical idea - you can't win without getting players on base. They're easy ideas, but because they challenged a comfortable status quo they've been resisted by a generation of columnists and ex-players whose income rests on being able to blow dog whistles (in the US) or spew cliches that their audience understands. So the debate is really about evidence vs assumptions, not 'stats' as such. I have a slightly hard time believing that Wells loves stats, because his success predates their current level of usage and I doubt accurate data is generated for the lower levels of junior footy he watches. But by the same token I think he clearly doesn't get caught up in the buzzword hype that surrounds a lot of players. He watches what they do on the field, talks to them (which I think he's said is the most important part of the process?) and looks for characteristics he views as critical.

Ramble away , you are one of the most informed and informative posters .

So basically the debate was "Moneyball" V "Trouble with the Curve" ? Statistical V Intrinsic ? Is Moneyball doing more with less , doing better with the over looked or under appreciated or is it finding new benchmarks to acknowledge talent

 

[Lana]

It's definitely happening - every year in the Prospectus CD stats guys talk about their latest effort to isolate the value of taggers, or teams scoring efficiency, or whatever. But they also stress that CD is a business and that as a result this analysis remains confidential so it retains value. I doubt journos get a look at it. Most amateur analytics in US and European sport also relies on freely available data that is a lot more sophisticated than what we get here.

It's coincidental to this article but its interesting that a number of the most prominent amateur analytics guys in North American hockey have been hired this offseason by NHL clubs. Just as in Moneyball (arguably more so) there's been a pretty overt war of words between traditionalists and stats guys in the NHL and the hiring of a bunch of bloggers is seen as vindication of the bloggers. But really, they already won - they've kept it quiet, but dominant teams of late like the Kings and Blackhawks pretty clearly spend big money on their own analytics.

That said (while Im rambling) I also reckon the debate in hockey shows why you have to be careful of getting carried away with stats in and of themselves. The debate in hockey boiled down to an argument over whether or not possessing the puck is important. The answer is pretty obviously yes - the team that is better able to hold onto the puck will shoot more and score more goals over the course of a season. In much the same way, on-base percentage is a pretty simple, logical idea - you can't win without getting players on base. They're easy ideas, but because they challenged a comfortable status quo they've been resisted by a generation of columnists and ex-players whose income rests on being able to blow dog whistles (in the US) or spew cliches that their audience understands. So the debate is really about evidence vs assumptions, not 'stats' as such. I have a slightly hard time believing that Wells loves stats, because his success predates their current level of usage and I doubt accurate data is generated for the lower levels of junior footy he watches. But by the same token I think he clearly doesn't get caught up in the buzzword hype that surrounds a lot of players. He watches what they do on the field, talks to them (which I think he's said is the most important part of the process?) and looks for characteristics he views as critical.

My Issue with an AFL version of "sabermetrics" being use for drafting is that you can't use the stats from the lower levels as a way to predict performance is upper level. The difference of talent,physical characteristics and tactics make some players look dominant at one level but would be ineffectual at the higher level.

It works in the USA because the jump between the second tier and top tier league would be smaller due to the larger pool of sportsmen and larger ease of mobility of players between teams and leagues. I wonder how the AFL would be different if the AFL could draw players from a significantly larger population. Many games are dictated by the fact that teams can't find a match ready player for certain positions.

 

[Lana]

Excellent example. I was thinking older guys who become good solid players rather than young kids , not long drafted which get people all hot and sticky. I made an example in regards to Frawley when some rated the comp as over the top.... Draft Picks , most of the time under deliver compared to a know commodity.
Im watching Chappy kick a goal right now. He costs Ess nothing gives them 15+ games , compare that to so many kids which take for ever and give little.
Its no wonder the bigger , smarter clubs can not ignore FA.

On you point of the shorter player , it will be interesting who calls Daniel out on Draft Day.

I am firm believer in the value of early picks, most of the time they fail, it isn't because of lack of talent, it's because they have a horrid run with injury, the player has an attitude problem or that there attributes simply don't fit in with the current tactics employed. Scouts generally do a good job. But list sizes at AFL clubs are massive, at least 44 players, I'd believe that you get more marginal benefit using a late pick on a mature age player than a kid who would be rated 150th on a list towards the end of the rookie draft.

 

[thejester]

Ramble away , you are one of the most informed and informative posters .

So basically the debate was "Moneyball" V "Trouble with the Curve" ? Statistical V Intrinsic ?

Aw shucks Turbo

Yeah the debate was basically about 'the eye test' vs 'stats'. The thing is, puck possession dominance in hockey doesn't automatically translate into wins because a variety of other factors can override or distort that possession. You might come up against a goalie who is playing out of his skin and so even though you're dominating, you just can't get the puck past him. Equally, you might just be unlucky - IIRC even the best players in the league generally score on about 10% of the shots they take, but that can fluctuate. So I guy might put up an amazing season for goals and you think 'he's finally reached his potential' when really hes just been shooting at 15%; equally, guys might be shooting at only 5% even if they're taking a ton of shots. Over the course of a season those things should even out, which is why puck possession is considered a valuable way of judging how teams will perform.

That puck possession doesn't automatically guarantee wins is the stick used to beat stats with in hockey, because teams with terrible possession numbers have gone on big winning streaks because of those other factors. Case in point are the Toronto Maple leafs: the biggest team in hockey (city of Melbourne's size, crazy about the game, with one team) that hasn't won since the 60s. They are terrible and they are terrible at puck possession, but they've also had some outstanding goaltending that has kept them afloat and delivered some big streaks. So a lot of the media abused stats-types when the Leafs were winning, trotting out the stereotypes of bloggers in their parents basement, and then looked really stupid when - as the bloggers predicted - the Leafs imploded and either missed the playoffs outright or collapsed in the first round. It doesn't help that in the case of hockey the stereotypes about jocks vs nerds have all sorts of additional historical baggage. The traditional way of playing hockey in Canada has emphasized strength and 'grit' and 'toughness', in the same way we love inside players. 'Outside' (ie skilled) guys have traditionally been seen as a European commodity and not tough, manly etc. So when a bunch of nerds come along and say 'hey the teams with skill are winning and are going to win in the long term and grit and toughness aren't very important or valuable' its like an assault on the national character.

So tl;dr, the Leafs might win 6 out of 10 games with 'grit' and a bunch of sports columnists and former players turn around and go BUT HOW CAN THE LEAFS BE TERRIBLE WHEN THEYRE WINNING STUPID NERDS?!, because its happening right in front of them. But the nerds (correctly) predict that over the course of an 80 game season they're not going to have a positive winning ratio.

Is Moneyball doing more with less , doing better with the over looked or under appreciated or is it finding new benchmarks to acknowledge talent

'Moneyball' specifically was about the second, but more broadly 'sabremetrics' or 'analytics' is about doing all of those things.

 

[Turbocat]

thejester , STUPID NERDS lol