Thursday, April 03, 2014

What took so long?

Bad Dad and I were discussing Michael Lewis' story about the discovery of how high frequency trading is used to rob the pension and mutual funds that hold your retirement savings.

Much of Wall Street is out performing damage control to explain how the situation is very complex and no one can explain how they are robbing the majority.  Actually, it's not that difficult to understand compared to the kind of mysteries that scientists routinely unravel.

The big mystery to me was why it took 18 months for Bradley Katsuyama's team to discover what was going on. It would have taken much less time if they had a physicist on their team.

Consider JILA professor and NIST scientist Judah Levine, aka the nation's "time lord".  Back in the early 1990s, he explained to me (a graduate student at the time), how, when delivering time standards over telephone lines (and, later, over the internet), he had to take into account the time it takes for the signal to get to the other end.  Is the phone line copper or fiber optic? He also helped develop time synchronization via satellite (Navstar and GPS)*.  This is in addition to his geophysics research determining the source of earthquakes and other underground motions.

If they had Judah on the team, he would have solved the mystery over lunch.
* The speed of light is only a constant in a vacuum.  The variation is small through the atmosphere and ionosphere.  But, over long distances, light is slowed just enough that you have to account for the slowing.  In fact, you also have to account for special relativity.

Fun cocktail party tidbit:

Why can you read about military satellite launches before they happen all over the internet?  Because there is no longer any point in trying to hide them.  To find out when a military satellite will be launched, one only has to try to book a hotel room near a launch range and note the price.

Prices spike whenever a launch is about to take place.  If hotel rooms are expensive and scarce, but the open launch manifest of nonmilitary launches shows nothing, then you know what must be happening.

You don't even have to call up the hotels anymore.  Your software robot can just query the hotel websites regularly.

Information leaks happen all the time, with real-world implications.


  1. I haven't read the book, but I've seen the 60 minutes review, and read a bunch of articles. I don't think that he is saying anything that wasn't obvious or evident (in short nothing new) and is mostly trying to sell himself as revealing that the markets are rigged, but the markets have always been rigged to some degree or another, and there is always that potential for manipulation. It seems naive to think otherwise.

  2. Well, insider trading isn't legal and skimming as an intermediary is. So that's a difference that can be remedied immediately.

    As for how brokerage houses rip off their clients in dark pools, clients can take their business to another brokerage.

    But there was no way to avoid this type of legal ripoff. So this is different.

  3. I read the article. A good telephony team might have figured that out too, had anyone thought to ask them. Very shady and unethical.
    We encounter the timing and delay issue in booking travel as well. Hotels and airlines adjust their rates based on real time demand. My ticket might be $100 another booked a few minutes later might be higher..or lower, as they adjust the rates. Our customers get very frustrated with the variability of the pricing.


    1. Exactly. I think this is part of the backlash against techies in the SFBA. The apartment pricing, jumping hundreds of $$$ overnight is just too unstable.

      If plane fares spike, leisure travelers stay home. Business travelers can negotiate rates on frequent routes. But, everyone has to live somewhere and moving is a huge hassle (if there is anywhere to move to). If people don't feel secure for a vital need like housing, then they can get very angry and lash out.

  4. NIST has a sweet new clock, too:

    accurate to a second in 300 million years.

    I've been peripherally involved with software support for the bank trading systems, can confirm they get very agitated about milliseconds..

  5. "It would have taken much less time if they had a physicist on their team."

    I can't speak for Brad Katsuyama's team, but I worked on various trading floors of large banks for 10 years or so, and quant teams are littered with physics PHD's. They're the polite ones who make all the money.

    1. Yup. That's what puzzled me. Why didn't they have a "rocket scientist" on their team?

    2. Maybe they could not afford one. Scientists who sell out and go into finance expect to be (and are) compensated well for betraying their own conscience.

    3. Um, just how much $ are we talking about? Did I tell the story of how I flunked my Wall Street interview when I was finishing up my PhD?

      The recruiter said that she sensed my values were too deep to be assuaged by a big salary. I never got to learn just how big a salary I lost with my lousy acting skills.