it seems that advances in video is making basketball analysis incredibly interesting. one would think that approach could also be applied to soccer and hockey.

also, my understanding is that hockey is regarded as the most random of the major sports in terms of outcomes. is that still accurate?

also, my understanding is that hockey is regarded as the most random of the major sports in terms of outcomes. is that still accurate?

Yes, (an no) because it (probably) has the greatest parity and also the inherent nature of the game makes it just short of baseball in terms of randomness. Baseball is more random on a game-by-game basis.

Hockey probably has the most room to grow from a management perspective, though, since the management landscape is still made up primarily of Canadian good old boys. The first European GM was just hired in the last month.

I think the more moving pieces you have the more difficult advanced metrics are to come about so I'd say soccer and hockey. Of course both sports have advanced metrics and there is some great work being done in both so it's not impossible.

I think tactics and team wide concepts are where low hanging fruit may exist in those sports.

Hockey and Football are the hardest (I am not sure which is harder. will think about it). Basketball is easier than hockey because the games are similar, but there is much more scoring (duh) which evens out the randomness. Baseball is the easiest because it is much more static and measurable than the others.

zack

the study i remember had hockey burying the needle in terms of game to game randomness. the other sports were not close. i will hunt the google and see if i can find it

Scuba diving.

Competitive Murray Chass in a Fish Barrel Snarking?

In baseball, all 30 teams are doing it—there’s no dent.

Somewhere in Florida Ruben Amaro Jr. is LOL-ing.

Pro wrestling (although Barry Horowitz was the living embodiment of replacement level).

STATS, INC owns the SportVu system that is generating a lot of that basketball data, and I know they are trying to get the system installed by NFL teams and soccer clubs. I personally think that we might need some new techniques to have really good analytics with hockey, basketball, and soccer. Those sports really require analysis in 4 dimensions.

The idea that there aren't still some pretty stupid teams in baseball seems naive. There's also a ton of room for very, very profitable research, IF you can find the answers, such as identifying the likely collapse indicators for 30-32 year old players...

Which sport is the most immune to Moneyball?

I think Interpol has demonstrated pretty conclusively that it ain't soccer.

The real advantage of baseball is that, taking fielding out of the equation, you're talking about one-on-one matchups. Without interaction from the other players, it's much easier to isolate an individual player's contribution. I think that's why basketball is the second most advanced of the sports... even though there is some interaction between players, especially in today's NBA, there's still a lot of isolation you can do on an individual player's performance. Hockey is much tougher because very little happens in isolation and players are actually on the ice for no more than 20-25 minutes a game (much of it spent doing very low event probability actions). Football seems like it would be the toughest because you have 11 guys on each side, most of which are contributing to the play, but away from the ball. How do you measure the effect of a receiver's blocking on a running game? Does it depend on the quality of their offensive line? Even measuring the contribution of the running back relative to his line is complex. And there are so few actually plays run in football (about 60-70 per team), it seems very difficult to sort out who is responsible for what in any given game.

Statistical analysis has close to no value in boxing.

Mixed martial arts is doing some things in trying to develop statistics, but they have a lot of obstacles- all "significant strikes" are not created equal.

I don't pretend to be an expert at any ML sport besides baseball. BUT.....

It seems obvious to me that MLB shoud have been the LEAST likely to need "Moneyballing". It's far MORE an individual sport than the others. It's far more OBVIOUS what batters are worth, becusse they largely don't interact with teammmates. OTOH, for EVERY other sport it should be much MORE challenging to deconstruct individual player values. How do we know how much offensive linemen contribute to the running game? How do we value free safeties? What do we measure in hockey besides goals and assists and goals/shots allowed and PM? How do you best quanitfy defensive ability on the hardcourt? I doubt there has been 1/10th of the ground uncovered in these areas versus the efort put in on beter run estimaors.

And of course, if I cared a whit about hockey, I wold do something about it...

Statistical analysis has close to no value in boxing.

Just what I was thinking. I mean, there are strategies, but no real statistics that can be measured impartially.

I think Interpol has demonstrated pretty conclusively that it ain't soccer.

I laughed, but that's also true I think. If you take moneyball, the strategy, from Moneyball, the subtitle: "The Art of Winning and Unfair Game", then soccer is by far the least "moneyballable". The best you could hope for is a few year run to a European place before your team is ripped apart. Not to say that there aren't teams that punch above their weight consistently, but success is a far more relative term in that context.

I think Interpol has demonstrated pretty conclusively that it ain't soccer.

Well played.

The idea that there aren't still some pretty stupid teams in baseball seems naive. There's also a ton of room for very, very profitable research, IF you can find the answers, such as identifying the likely collapse indicators for 30-32 year old players...

I also think the notion that we've come to the end for "moneyball" in baseball is naive. Pitching staff usage is still the same as it was in the late 80s. Roster compositions are still as they were 20 years ago. You still see speedsters hitting at the top of lineups and sluggers hitting cleanup. We don't fully know the effects a manager has on a club. I don't think we know everything about the impact of defense and baserunning quite yet. The returns are diminishing, but there are still edges to be had in baseball.

Mixed martial arts is doing some things in trying to develop statistics, but they have a lot of obstacles- all "significant strikes" are not created equal.

Like boxing, the real value comes from reviewing video footage. Quantification adds nothing. If I know a guy likes to lead with a left hook, step in, and fire a right uppercut, knowing he does this after 71% of his lead left hooks doesn't provide any tangible value.

I'd think soccer's going to be the hardest because of how subjective what qualifies as a "scoring chance" is going to be. I suppose you could quantify who wins midfield battles, who's good at controlling passes, and other useful things though. It's just that the punch line is soccer tends to be more artistry and luck than easily statistically quantifiable events.

Football's also going to be hard to quantify because it's not always clear whose responsibility is what on a play, even with video to slow things down and watch them multiple times. How do you know if the receiver missed the route or if the QB threw to the wrong spot? How do you know if the running back was supposed to stay in and pick up the blitzer? How do you know which DB had responsibility for a certain spot in a complex zone scheme? None of these things are clear at all on quite a few plays, without knowing what the scheme was.

Hockey's basically basketball with less scoring and more randomness, for analytic purposes.

Pro wrestling (although Barry Horowitz was the living embodiment of replacement level).

Don't tell Bodydonna Skip.

We're not too far removed from an England manager who capped a player he'd never seen, and another who favored psychics. I've seen no-shooting goons favored over 'fox-in-the-boxes' and guys who could skin players but couldn't pass for #### favored over those who could.

I should not be surprised at all to find Annie Savoy's bit about 'fallen fruit no one sees' to be much more applicable to soccer than baseball. It's not hard to imagine soccer players with amazing vision in taking passing lanes passed over cause they were shite at playing keepy-uppy.

edit: Also, home and away results are so dramatic, that they're included in tournament rules. I'm not sure what you can do to get players to ignore that (or refs) but it would be a handy ability.

We're not too far removed from an England manager who capped a player he'd never seen, and another who favored psychics.

And neither of those would rank in the top 20 for 'crazy #### Raymond Domenech did while manager of France'.

Like boxing, the real value comes from reviewing video footage. Quantification adds nothing. If I know a guy likes to lead with a left hook, step in, and fire a right uppercut, knowing he does this after 71% of his lead left hooks doesn't provide any tangible value.

If nothing else, that kind of stuff is interesting for outside observers, so it's not totally worthless.

I do agree that sports with discrete individual actions, like boxing and MMA, are less likely to have exploitable inefficiencies than team sports.

We're not too far removed from an England manager who capped a player he'd never seen...

He shot him? Jesus. What provoked it? Did he die?

Pitching staff usage is still the same as it was in the late 80s.

Demonstrably untrue. Starting pitcher usage, and relief pitcher usage beyond the closer, continued to evolve until the late 90s.

-- MWE

I should not be surprised at all to find Annie Savoy's bit about 'fallen fruit no one sees' to be much more applicable to soccer than baseball. It's not hard to imagine soccer players with amazing vision in taking passing lanes passed over cause they were shite at playing keepy-uppy.

But in soccer they are going to get a chance somewhere. There are hundreds of pro teams looking for talent. In baseball teams keep guys they don't believe in around, it costs peanuts to keep them in the minor leagues. It's way easier to wither on the wine in baseball, a soccer player may have to go to Sweden*, but he will get to play competitive games somewhere.

*) And a small soccer league is nothing like a baseball minor league. A player is free to negotiate his own terms.

Mike Emeigh is correct. I think we've gone from a couple hundred >130-pitch starts a year 20 years ago to ten.

Like boxing, the real value comes from reviewing video footage. Quantification adds nothing. If I know a guy likes to lead with a left hook, step in, and fire a right uppercut, knowing he does this after 71% of his lead left hooks doesn't provide any tangible value.

I disagree; if you could give me numbers on not only how often my opponent does something, but how effective he is with it, I think it makes me more able to fight him. Maybe this is because I don't know jack about boxing, but we do this sort of quantification in basketball (Player X goes to his left Y% of the time and shoots Z% going that way, so shade him that direction). Maybe it's different because a boxing match ends if you guess wrong or your opponent bucks their tendency, as where a basketball player just gets a single bucket.

Still, quantifying boxing tendencies seems useful to me.

I disagree; if you could give me numbers on not only how often my opponent does something, but how effective he is with it,

The boxing analysts should use Hit Points as the statistic to watch.

Football's also going to be hard to quantify because it's not always clear whose responsibility is what on a play, even with video to slow things down and watch them multiple times. How do you know if the receiver missed the route or if the QB threw to the wrong spot? How do you know if the running back was supposed to stay in and pick up the blitzer? How do you know which DB had responsibility for a certain spot in a complex zone scheme? None of these things are clear at all on quite a few plays, without knowing what the scheme was.


OK, but your last phrase there is the key. Baseball analysis really blossomed with people outside the game. The numbers are mostly available to everyone, so outsiders could really go at it. In football, there may be the same opportunities, but they will likely only be available to the insiders. And more than that, they will end up being unique to and individual team, and for that matter, they may only be valid under the current coaching scheme. So you bring in a new coach, throw out all of your analysis.

Even the insiders may not be able to analyze players on other teams properly. Although maybe you could find a guy who would work well in your scheme but is being underused by his own team.

It's way easier to wither on the wine in baseball, a soccer player may have to go to Sweden*,

Well that's true...if he can overcome any bias attributed to his home country.

and there's another inefficiency. How many really talented Americans fall to the wayside cause they'd rather make 100K a year than 250 Euros a week? And...

Serious question: DO the major soccer powers even have scouts in the US going around to see the top high school players?

OK, but your last phrase there is the key. Baseball analysis really blossomed with people outside the game. The numbers are mostly available to everyone, so outsiders could really go at it. In football, there may be the same opportunities, but they will likely only be available to the insiders.

This is a great point. For some bizarre reason, the NFL refuses to release or sell its overhead camera shots that show all 22 players at once.

...What about bowling?

Baseball analysis really blossomed with people outside the game. The numbers are mostly available to everyone, so outsiders could really go at it. In football, there may be the same opportunities, but they will likely only be available to the insiders.

Dean Oliver was talking about this on HUAL recently (Oliver has been called by some the Bill James of basketball, though the comp is unfair to both). IIRC, he estimated (guestimated?) that this retarded growth in metrics by 30% or so. The next major frontier in hoops is the SportVU data - it's going to be interesting to see how teams (and the league) handle working with that data (I linked to on an article from the VORP blog last night that touched on that task ... data management becomes more relatively more important compared to manipulating end products).

...What about bowling?

What would be the advantage? It would have to be a player training tool, because there are no teams trying to acquire talent.

Same for boxing, tenni,s and golf. When it's just individuals playing for themselves, there's no one with an economic interest in investing in analytics.

Except maybe gamblers, and that work won't be made public.

Here's an analytical basketball question...is 'hot hand' real and how much should it affect the game plan?

Serious question: DO the major soccer powers even have scouts in the US going around to see the top high school players?

Yep, only they aren't watching High School soccer usually. Club soccer for the most part.

The bigger problem is that as a country we're not very good at the sport, and we're even worse at coaching it. And so ultimately the European scouts don't come away with much. There's also a protectionist rule in place that forbids teams from signing foreign players until their 18th birthday. This rule exists because it's effectively a subsidy to club teams in countries without the big leagues.

But a Junior Flores gets snapped every now and then anyway.

For some bizarre reason, the NFL refuses to release or sell its overhead camera shots that show all 22 players at once.

Baseball has some weird ideas about who gets access to some of the newer data too. I understand SportVision has to get paid, but it's not clear how they benefit when they keep their data secret.

What about bowling?

Maybe the pros don't need it, but I would love to know if there are certain paths that are strike-happy or split-prone.

This is a great point. For some bizarre reason, the NFL refuses to release or sell its overhead camera shots that show all 22 players at once.

They finally changed this. They now offer NFL Game Rewind, which includes the "all 22" video.

Even the insiders may not be able to analyze players on other teams properly. Although maybe you could find a guy who would work well in your scheme but is being underused by his own team.

As a Patriots fan, I've gotten to watch a classic example of this, a nifty return man and backup WR who started only 3 games in 3 yr for the Dolphins, now averaging 112 catches per year for the Pats. (Had 96 catches total for his 'Phins years.) Of course, having Brady throwing you the ball rather than "Dolphins QB" also helps.

Serious question: DO the major soccer powers even have scouts in the US going around to see the top high school players?

They have extensive enough networks that they know which 16-17 yo Swedish players to invite for trials. I guess they mostly use local guys on a small retainer that research players, and then goes to see them.

Most young players are probably moved by their agents who markets them and gets them trials. Getting an agent isn't a problem for a reasonably good player, they swarm. The problem is finding a not-too-dishonest one.

38 - evidence suggests 'no'
I think it's more complicated than that, but there you go.

Here's an analytical basketball question...is 'hot hand' real and how much should it affect the game plan?

As expressed in #44, I'm under the impression that the basic answer is "no". Players probably can't really go on streaks that are anything particularly out of their expected deviation from their true talent level. That said, players can believe they're on such streaks and decide to shoot more, which makes it sensible to shade a defense in their direction if that behavior is observed.

As an observer and player though, it's hard for me to believe the statistical evidence. I think the lack of apparently "legitimate" streaks might be a result of exactly that sort of defensive adjustment being made. When Lebron James seemed to have gone to an NBA Jam level He's On Fire situation against the Celtics in Game 6 last year, it'd be pretty hard to convince anyone watching that Boston should have just played him normally and he'd "cool down".

There's also the issue of process v outcome. I remember a game where I was the hot shooter on my team ... reason was that the guy guarding me kept ignoring that I was lefty with a quick first step. I wasn't hot - I was more open than normal. Also, hot is not the mirror opposite of cold - cold can mean something's wrong, hot usually means something isn't and you're lucky.
There was a recent paper supporting the hot hand hypothesis - but I think that, too, was measuring something else...

I think football is in a pretty good spot. You probably can't analyze every player, but you can analyze the overall strategy and make some significant changes if you have the guts. There have been studies on the validity of the punt strategy, and some coaches have seen interested in changing that approach, but just don't have the guts in the end. When a team decides that playing for four downs instead of three, it's going to have a ripple effect throughout the league. Of course the first team to do that, is going to have to have a strong defense to get away with it.

Football is still pretty far from being over run with statistical analysis, that it's in it's infant stage. At this point in time, any advances you get from analytics will be on the coaching side, and not having to do as much with the individual players.

...What about bowling?

Videotape is probably the best tool for improving your bowling, Although for an amateur, having stats on spare pick percentage and types could be useful to improving your game. You might think you make a lot of a particular spare, and then you find out that you are only 70% or so on that spare, and you might concentrate more. I know that there are are few single pin spares that I used to be cocky on, because they were so easy, then I started to pay attention, and my lack of effort on them had led to a lower than expected percentage of picks.

Bowling has the same problem as any other sport though, if you just go by your perception of how you are doing, you might fall into traps. Bowling would be among the easiest sports to quantify.

For football, and perhaps hockey, there should be an opportunity to use slo-mo video and calculate a distribution of "time to react" and "time to get from X to Y" in a fashion that might be useful. I don't doubt that the coaches already know which players are "quick", that is, have a low reaction time, but how does the distribution of those times change depending on variables such as: how many plays has he executed today, day/night game, what player opposes him?
Really baseball should be quantifying these measures as well. A lot of people here recognize fielding metrics as needing work.

it'd be pretty hard to convince anyone watching that Boston should have just played him normally and he'd "cool down".

The correct answer, as to when you should play LeBron 'normally', is 'never'.

As expressed in #44, I'm under the impression that the basic answer is "no". Players probably can't really go on streaks that are anything particularly out of their expected deviation from their true talent level. That said, players can believe they're on such streaks and decide to shoot more, which makes it sensible to shade a defense in their direction if that behavior is observed.

I believe the study done found that after a certain number of makes players overestimated their abilities and took worse shots, meaning having a hot hand actually hurts at some point.

As an aside, Carmelo Anthony has said he does this - he calls them 'heat checks'. He takes a shot he normally wouldn't, a few feet behind the three point arc early in the shot clock maybe, just to see how much he is feeling it. Hard to argue with him too much though, when he is on he is pretty unstoppable.