flourpower

I couldn't find this by googling - if A is an n by n matrix, can you get A^k strictly faster than you can get a product of k arbitrary n by n matrices?

This is unfair. Some of his work is built around general relativity:

http://www.princeton.edu/~pkrugman/interstellar.pdf

You're thinking of CDS, not CDO. CDS stands for credit default swap(s) and CDO stands for collateralized debt obligation. You're right that a CDO would be really big if you printed out a formal specification, but that's because a CDO is special trust where the trustee buys and sells different securitized products and tranches out the payments to shareholders in the trust.

Also - AIG could have made the same mistake with conventional home insurance. Say they only keep 100 dollars of cash around and they decide to insure a million houses, each with a value of one dollar. If one ten thousandth of the houses burn down, AIG goes bankrupt. So it's not true that selling a dollar notional of home insurance is less risky than selling a dollar notional of CDS, because it could easily be the case that the expected payout on the CDS is higher. CDS are just harder to price. There are very robust statistics about houses burning down - the statistics on whether homeowners would default were a lot trickier to deal with.

You're right, of course, but my point (and maybe I didn't make it very effectively) was that to acquire a similarly thorough understanding of macros would require way more information - I wasn't trying to make the point that a complete description of CDS could be given in a couple of paragraphs.

I think this tendency to mythologize certain financial derivatives is weird. The concept of a macro is way more complex than the concept of a credit default swap. To demonstrate, here is (in my opinion) an explanation of credit default swaps that a person with no financial background should be able to understand.

A bond is a contract created and sold by an "issuer". The issuer can be a government or a company. Ownership of the bond entitles you to payments from the issuer. The specific number and size of payments varies from bond to bond. Once the issuer sells the bond to someone, that person can sell the bond to anyone else they want.

A credit default swap is a contract between two parties (neither of which is necessarily the aforementioned issuer) that are usually called the protection buyer and the protection seller. This contract is made in reference to someone called the reference entity. The protection buyer agrees to make a series of payments to the protection seller in exchange for the protection seller's promise that, in the case of a "credit event", they will give the protection buyer either some specified amount of money or some specified amount of bonds. The nature of the payments and the meaning of "credit event" vary from contract to contract, but generally a "credit event" is understood to have occurred if the reference entity (which is always an issuer) fails to make payments on some of the bonds it has sold. Either party in the contract is free to find someone else to take up their side of the trade at whatever price they can negotiate.

The only thing you needed to know to understand that was what a contract was, but if you want to understand macros, you really need to know what an interpreter is.

I agree with your advice - I just don't think most people follow financial news that closely.

However, the euro is down about 8% since June, so it's safe to say there's been a fair bit of selling.

I think that in this case the reporter doesn't understand what's going on behind the abstraction.

One thing that might happen is that some country's equivalent of our treasury dept will realize that they're not going to be able to make the next set of payments on their bonds.

But then why not just change your euros into dollars?

Not that I have a solution to this problem, but who does the filtering once you remove the job offer as a prerequisite? I can't conceive of a situation where the people that get put in charge of that process are actually qualified.

That's more of a reflection of liquidity than of creditworthiness though.

That's the same as the probability that 2N flips of a fair coin result in exactly heads: (2N choose N)(.5^N)(.5^N).

Check out: http://en.wikipedia.org/wiki/Binomial_distribution

What Taleb is saying is still true. If I have a 400k market salary and I give up 200k of it for the right to 1% of the profits I generate, then I still maximize the expected value of my compensation that year by maximizing the size of my bets. I could bet a billion dollars on a coin flip, get 9.8 million (after recouping foregone salary) on heads and lose 200k on tails. Moreover, there is a well established history of traders that lost large amounts of money finding gainful employment regardless. See Boaz Weinstein, for example.

And even then it's tricky, because a non-regulated entity can make a bet with a regulated one and end up getting bail-out money by virtue of that bet. That seems undesirable, but you also can't really stiff the non-regulated entity, because then nobody will ever want to bet with the regulated one again, thereby destroying its ability to hedge.

But it can only go as far below 0 as the amount in salary you sacrificed in exchange for a potential bonus.

That would work if we wanted to stop banks from transferring losses to shareholders - the real problem is that they transfer losses to taxpayers. Pre-IPO Goldman Sachs was probably considered "systemically important" enough for their losses to have been covered in the event of a large trading loss.

3 would necessarily raise the bid/ask. If you charge everyone 10 cents to transact, then nobody will act as a market maker at a bid/ask narrower than 20 cents.

The prices now will be less artificial, since more people will be involved in setting them.

But Harvard provides incremental value over Khan academy by serving as a networking tool and a credentialing mechanism.

We shouldn't practice financial regulation like it's holistic medicine. Metaphors about cleaning the pipes and taking the pain should be weighed against the evidence we have about what happens when you let banks fail, and that didn't go very well when we did it with Lehman Brothers.

The level of uninformed fear surrounding this topic doesn't bode well for driverless cars. Computers trading stocks can't even kill people.

You're talking about volume, not price. Increased volume should, if a market is performing "properly", accompany new information. If there's new information, there's a reason to trade. The examples you cited seem like evidence of the stock market's predictive value - for instance, increased volume in Citi stock was a piece of evidence that something about the security was expected to change.

Moreover, it's not true to say that stat-arb guys don't care about future earnings growth - they just use statistical methods to project it. To use a simplified example, a stat-arb guy might automatically buy shares in some small-cap automotive supplier if Ford rapidly increases in price. If the increase in Ford stock represents positive fundamental information about the auto industry, they've applied that information to the price of the supplier faster than a human would have and they'd make a profit. If, on the other hand, it represents concern about some scandal involving the Ford CEO, they've contributed to the noise and lost money.

Isn't the claim that algorithmic trading will never replace human decision making basically equivalent to the claim that humans will never construct strong AI?

What evidence do you have that the stock prices have recently become less predictive of future earnings? Was the market performing its predictive duty in, for instance, September of 1929?

I think he should have treated the set of all government actions with more granularity. Imagine that you partitioned that set into two subsets. The first subset would be actions on policy issues that libertarians find it necessary for governments to act on. The second subset would be actions on policy issues that they don't think it's necessary for governments to act on. It could simultaneously be true that 99% of actions in the second subset will tend to have bad results and only 1% of actions in the first subset will have bad results depending on the relative size of the sets. If you're not clear about which subset you're talking about, you'll get people saying things like "you must be wrong that most government policies have a bad results, this country (that mostly performs actions in the first subset) tends to have very good results."

If you buy 100 dollars of mortgage bonds at a price of p and I buy it from you at a price of q, then your profit in that transaction is 100*(q-p), not 100. Besides, if you look at the profits made by investment banks since 2008 you'll see that they're nowhere close to trillions. Even the most profitable investment bank in the world, Goldman Sachs, had a total net income in the last three years of less than 25 billion. But your implication was that they made trillions as a result of what happened in the financial crisis, which is an even stronger claim.

Trillions? Even if you were referring to the amount of money made by all American investment banks in the last several years, you'd still be off by a few orders of magnitude.

Even after excluding a big mortgage related charge, Bank of America's net income last year was only 756 million dollars. If you estimated that each of the named banks would have to pay out an eighth of the 20 billion with the other ten billion dollars spread out across unnamed banks, you'd still be charging them 3 times what they make in a year. That's more than a slap on the wrist.