Slashdot Mirror
World Cup Forecasting Challenge For Quants
databuff writes "As a break from projecting the strength of subprime mortgages, credit default swaps, and other obscure financial instruments, quantitative analysts at Goldman Sachs, JP Morgan, UBS, and Danske Bank have modeled the 2010 FIFA World Cup. Now Kaggle has set up a forecasting competition, allowing statisticians to go head-to-head with these corporate giants. The challenge is to predict how far each country will progress in the tournament."
For somebody falling to the ground clutching their leg long enough to get a card thrown?
There are numerous problems here.
Firstly is very few fans of football can truly consider themselves independent enough to do this well. I will try, but I just know my bias's will in the end have some effect on the outcome of my selections.
Secondly it isn't just about stats at something like the world cup where there are very few second chances and It is a game where you can completely dominate the opposition and still lose to a single error or bad ref decision.
I wouldn't put too much stock in their predictions. You know its fishy when they forecast Angola as a "team with strong fundamentals and underlying security". (I'm not even sure Angola has a team).
Read what I mean, not what I wrote.
we at slashdot would appreciate it if you showed those big banks how to work properly with stats.
Does having a witty signature really indicate normality?
Informationally Partitioned Simian Launched Projectile Selection.
If you were blocking sigs, you wouldn't have to read this.
I mean, the financial market is still a mess and I'd rather have them working on the real issues we face. Or is this a quick glance at what actually always goes on at those companies? Are they nothing more than professional gamblers and don't care about their responsabilities?
For the record, quants rarely try to predict things in the market. That's left to people who work in econometrics. The main job that a quant does is to price financial instruments in a way that is consistent with the market prices of other liquidly traded assets. I'm being deliberately vague about what precisely is meant by "consistent" because that often depends on the choice of model, but there are also model-free results which require certain asset prices to obey certain relations: put-call parity, for example.
Slashdot: news for Apple. Stuff that Apple.
You're not as clever as you think you are.
<spam>If you think you're good at this sort of thing, you might want to join the free online prediction game I run. There's a US$50k prize up for grabs, if you're better than these guys...</spam>
(Yeah it's spammy, but check my account ID - it's not like I just signed up recently or anything)
Paranoia isn't an infectious condition, it's a way of life
How is calculating odds for gambling/sports forecasting any different from calculating odds for gambling/the stock market?
Just go to Vegas or look up the current World Cup odds at any online sportsbook and it is very likely that these odds are better predictions than whatever they came up. These sportsbook odds are based on actual money bet by people who know something about sports betting instead of financial types. If these quants were actually confident that they had some sort of an edge, they would put their money where their mouths are (and affect these sportsbook odds). Of course, these sportsbook odds are also affected by amateurs, but the same is true with almost anything else, like the stock market. Another issue is the smaller liquidity compared to the stock market, but the amount of money bet on the World Cup should be large enough to make it less of an issue.
For just $10,000 in unmarked and nonsequential bills, Vinnie "the kneecap" is willing to venture a prediction as to when any particular team is going to drop out.
"When yous got a problem, pick Vinnie. We don't predict; we Promise."
I've written an online Markov Chain Monte Carlo simulator where people can experiment with pairwise team predictions in order to not just predict tournament winners, but more interestingly, to evaluate the most profitable expected payoffs from current betting odds (which are almost never the most likely teams to win). The page goes into the statistical basis of Markov Chain Monte Carlo (MCMC) quite a bit, and the simulation is all done client-side using javascript. Might be interesting for those who like this kind of stuff. http://www.donationcoder.com/wcp
"Are they nothing more than professional gamblers and don't care about their responsabilities"
Stock markets are nothing more than casinos for the more "respectable" end of the social spectrum.
Spain or Brazil - the same as the bookies. So what do the brokers say?
Banks are using other peoples money to make money for themselves, then claiming it's all down to such highly skilled workers (what a joke).
Brains / intelligence has NOTHING to do with the amount of money you earn. I mean am I really supposed to believe that a nurse that wipes other peoples dirty asses are worth less than a gambler? Or a surgeon that saves lives?
Banks that use other peoples money SHOULD be paying dividends back, not cashing in and paying their bosses six figure salaries. It's not as if a clerk gets a share of that whopping bonus pot. Oh and has anyone actually been charged with fraud yet because that is exactly what happened, on a grand scale!
Why hire Vinnie when the locals are so much better at it ?.
This is a classic case where the mean crowd response is likely to have greatest predictive success - in which case just looking at global betting markets will give the most accurate prediction.
Are they releasing these "odds" to the public while betting against them?
Who knew JP Morgan had a sense of humor? I mean, England World Champions? Hilarious!
"You can't allow somebody to commit the crime before you detain them." [Condoleezza Rice]
15 years ago, when I was working as a Systems Analyst at a Brazilian bank that shall remain nameless, it was common knowledge that trading desks all over the country were engaging in this kind of thing. They would create "financial products" tied to World Cup statistics and use all the technology, corporate and individual knowledge at their disposal to try to predict the outcomes and win or lose huge sums of money. Individuals bet with their own money and the corporations they worked for (and who provided the infrastructure for this) tended to look the other way. One such "product" I remember well was the so-called "GDC" ("Gols Da Copa" - Cup Goals) which created a market around the total number of goals to be scored during the World Cup. I knew one trader who payed for his house with his GDC money. Most of the time it was a mostly harmless hobby (if you discount the fact that gambling is illegal in Brazil) but as the World Cup final approached, I was very aware that the resources who were supposed to be working on models of commodities, foreign exchange and other markets did little else than model the World Cup; this included both people and computational resources. I wonder if some of my old colleagues in Brazilian banking ended up finding positions in Wall St.
For American non-football fans, the "World" in "World Cup" means that lots of different countries from around the world participate. Different kind of "World" from "World Series Baseball" which I believe has a different interpretation of what the word means ;-)
Sorry, couldn't resist it ;-) Hey, you're in the football world cup too, and you're not too bad at the game either!
(yes I know it might just mean the name of a newspaper rather than a particularly limited view of how many countries there are out there...)
What is "The country where pizza was invented, Alex?"
...the future crusty old bastards are already drinking the Kool-Aid.
http://bit.ly/a5OAnH
(usenet, rec.sport.soccer).