Wired article on copulas
Reader Scott Bury forwarded this article from Wired on the probability function that many used to quantify the risk of certain financial instruments. You can think of a copula as a way to represent multi-variate probability distributions. Nothing wrong with them. But here was a case where everybody used a model because everybody used the model but nobody bothered to check if it worked. We know the results.
My poor Detroit
Nolan Finley of the Detroit News writes of rampant racism on Detroit’s city council. Anybody have Al Sharpton’s number?
A pitiful Teamster official who practically crawled to the table on his knees expressing profuse respect for this disrespectful body was battered by both the crowd and the council…
Desperate, he invoked President Barack Obama’s message of unity and was angrily warned, “Don’t you say his name here.”
Even Keynes didn’t agree with Keynesian economics
[T]o believe it is possible for governments to spend our way to prosperity would be a major error in policy. There is no previous occasion in which such spending has been shown to work, while there are plenty of instances in which it has not. On every occasion that such spending has been used, the result has been a worsening of economic conditions, not an improvement.
Article here.
Global warming “on hold” for 20, no, 30 years—maybe more
Everybody knows it hasn’t been getting warmer, but it has been getting colder. So, are the climate models which predicted warmer and not colder wrong?
Well, Kyle Swanson of the University of Wisconsin-Milwaukee says, “temperatures should have gone up”. But nobody told the temperatures. Isaac Held of the National Oceanic and Atmospheric Administration in Princeton said, “warming might possibly slow down or even stagnate for a few years before rapid warming commences again.” Not to worry, “we’ll have explosive warming” Held says.
In other words, the models have been wrong up to now, they are still wrong, but we still believe them.
(Article here.)
Genes and chess
Reader C.B. sends this article which purports to report “Male superiority at chess and science cannot be explained by statistical sampling argument.”
Some people think men are better at chess because men are stacked at the top of all ranking lists. But the folks at Gene Expression don’t like a paper that offers this argument: the paper argues men are on top because men either like or are encouraged to play more chess. Could be true.
Only way to rigorously test sex-ability differences is to take two groups of infants, boys and girls, raise them by robots isolated from all society, then train and test them for differences.
Not rigorous is using history: where there’s evidence of a big difference (like chess), your only argument against an innate difference is circumstance or discrimination. Both arguments are of course weakened when the circumstances are highly varied and discrimination non-existent. No fair using biology.
I live in the Washington DC area and Monday demonstrators were protesting global warming. We had six inches of snow Sunday night. The pictures of the demonstrators out in the snow were so funny.
Men are at the top of competitive Scrabble as well as chess, and it has been my experience that far more women play Scrabble than men.
Chess is for boys, and they can keep it. They care more than we do who wins.
My father has beaten a grand master (in a simultaneous) in his hay day and still plays to high level. He had two daughters that were sent to a junior school that had a chess fanatic as adeputy head and who trained all the children to play.
My first and only appearance in a match at age about seven involved me being beaten by a boy of the same age that was so disgusted at my foolish moves that he sat shaking his head and tutting through the whole game like an old man, the cheat! He probably fancied me really.
Psyched me out and I lost in just a few moves. Never played again until I was eighteen, I played a totally blind boyfriend who found my moves equally entertaining. I’m great at Othelo!
My Nephew has inherited the ability, I think, if only he could sit still for long enough.
So, being of chess genes, being taught from a young age, but not encouraged did not produce a good chess player.
In contrast, my father was recently beaten by an eight year old girl, the word hustled was used. She is a new member of the club. My father was amazed and charmed, not sure if he had a rematch, could have been too embarrassing. She might have let him win.
So, by using that particular probability function, they’ve copulated our economy?
(Sorry, but someone had to say it.)
Kevin,
Yes, but in a reversal of the normal ending of these kinds of events, we’re being asked to pay them for it!
Don’t you think asked is rather a polite way of putting it?
If you hadn’t said it Kevin, I would have.
That Wired article was fascinating. Couple of questions:
– Is The Journal of Fixed Income a peer reviewed journal? I can’t tell from the web-site.
– Did Li’s article spawn all manner of interesting peer reviewed science? I’ll bet it did!
– And why would the fact that ‘…the outputs came from “black box” computer models and were hard to subject to a commonsense smell test.’ be of interest to anybody? Isn’t the fact that something has been programmed into a computer and that the program works for a few years proof of its correctness to make policy decisions for at least 100 years?
Just wondering.
One more thing…
How did we ever let our economy become so dependent on a mathematical theory based on “Sklar’s Theorem.” Sklar is obviously the name of a barbarian prince from another planet sent here to destroy us all.
Joe,
Yes, it is peer reviewed, and, yes, Li’s work is referenced by other peer-reviewed articles. But there’s nothing wrong with Li’s math, you understand. He did not prove something that was false. It is that people took something that was true and used it where it was not applicable.
The last answer is no (as you know). The model essentially spit out a price, which was used. In that sense it worked. But the model also quantified risk, which was the more fundamental operation, and that was not checked. A risk contains a probabilistic forecast, and there are methods of checking these kinds of predictions with actual outcomes. To do so you have to be very careful to explicitly define the probability and outcomes. No ambiguity is allowed.
You must set up a routine of checking models. I know that was not in place because I have been trying to sell that idea to people for years. But nobody has, until now?, seen that necessity.
Nice article about Keynes. I only wish Keynes was the source of the above quote. It is instead from the author of the article.
Joel,
Right, I should have made that clear. I should have put up other quotes from Keynes.
I think you are misreading the commentary at the linked Gene Expression post. It disparages the paper that can be found here:
http://bura.brunel.ac.uk/bitstream/2438/2967/1/Bilalic+et+al+–+PocB-final.pdf
The paper states in its abstract:
“Although the performance of the 100 best German male chess players is better than that of the 100 best German women, we show that 96% of the observed difference would be expected given the much greater number of men who play chess. There is little left for biological or cultural explanations to account for.”
The commentator at Gene Expression uses the whole post to take apart this argument. I believe it is the original paper, not the Gene Expression post, you mean to disparage.
“It is that people took something that was true and used it where it was not applicable.”
Math meets engineering.
Let’s hope that this is the last time a group makes that particular multi-trillion dollar mistake with the economy for at least let’s say, oh, 2 years.
Curt,
Oops. You got it.