Neutrinos to Give High-Frequency Traders the Millisecond Edge

Recently US plans for the LBNE next-generation neutrino experiment have run into trouble finding room in projected HEP budgets. Today (via Emanuel Derman’s twitter feed), I learn of a promising new source of funding. A Forbes columnist reports here on prospects for using neutrino-based communication through the earth to do high-frequency trading, arbitraging prices in markets on opposite sides of the globe.

To actually do this, I’d guess that financial firms would have to site machines like Fermilab’s proposed Project X and detectors like the LBNE one close to the servers running the markets. When they do this, maybe they’ll let physicists use them on weekends to do physics. It has been unclear whether the US government could afford to build Project X/LBNE, but surely Goldman Sachs and other major investment firms would have no problem coming up with the billions needed. Yes, I’m well aware that this is a completely insane and ridiculous idea, but that hasn’t been an obstruction to Wall Street innovations in recent years.


Update
: For the latest on LBNE, see this from Nature, out today.

This entry was posted in Experimental HEP News. Bookmark the permalink.

18 Responses to Neutrinos to Give High-Frequency Traders the Millisecond Edge

  1. Michael Hutchings says:

    Was this article posted exactly one month late by accident?

  2. Peter Woit says:

    Michael Hutchings,

    No, but it shows the problem one has trying to come up with April 1 postings. Truth is stranger than fiction. Actually, in this case I’m pretty sure this idea is unrealistic. But, I may be wrong…

  3. Edward says:

    Markets on the opposite side of the world are typically not open at the same time, which makes this less useful than it might be.

  4. paddy says:

    Let us calculate the bit rate available per Watt…..oops..never mind.

  5. Yatima says:

    I don’t understand. Why would the computers (seller or buyer and the exchange) be on opposite sides of the globe so that the need to communicate with each other with low latency becomes a concern? Typically, one would set up seller and buyer computers in the same housing company building as the exchange computer. No other communication would be needed or useful, so there will just be a couple of routers and firewalls and a few metre of twisted pair cabling to pass. Maybe I’m being naive.

    Off-topic, but there recently was this fun article on HFT.

  6. Peter Woit says:

    Yatima,

    I think the point is that you want to be making a decision to buy or sell based on information that is 44 milliseconds ahead of your competitors trying to make the same decision (since your information channel goes straight at the speed of light, theirs goes around the globe). Of course, once everybody installs an accelerator, neutrino beam and neutrino detector in their office building, the things would kind of become worthless…

  7. Anonyrat says:

    The time to modulate a neutrino signal (even just on-off) is probably longer than the difference between the light travel time around the globe and through the globe.

  8. anonymous says:

    Maybe a few physicists who work on the project will get royalty contracts and become extremely wealthy (another quant bubble? Sign me up!). Wall Street could use some scientific involvement, though – maybe they could even learn the real calculus instead of that watered-down version they have to take.

  9. Z says:

    There was a March Physics World article on neutrino communication: http://physicsworld.com/cws/article/news/2012/mar/19/neutrino-based-communication-is-a-first

    There is also another, military application, for covert communication to submarines.

    The gist of it is that with current detector design and error correction techniques, the bit rate is about 0.1 bits/sec for a neutrino beam. This is a far cry from even the 160 bit/s that Voyager had, which is a celebrated example in information theory.

  10. S says:

    I think it’s pretty clear they’re hoping FTL neutrinos were true. Think of the money they could make then!

  11. M. Wang says:

    Two years ago, when a new firm raised $300m to dig a STRAIGHT-line trench for a new fibre optics line between Chicago and New York, I had joked with a colleague that the next step would be neutrino beams, which was the only thing capable of avoiding the earth’s curvature. I guess someone else has now come up with the same idea, but not as a joke.

    The reason why these investment makes sense (to varying degrees) is the physical separation of trading activities in two sets of financial assets: stocks and the stock-index-based futures. The former trades in New York, but the latter trades in Chicago. Usually, the big pension funds or mutual funds find it cheaper to enter/exit a big position by buying/selling futures than the corresponding stocks. Therefore, the futures price would move first. The competition to use this information to make money on the slower stock markets is intense. The aforementioned new fibre line saves 2ms in communication time, for which the firm charges $1m per month. Only the leading arbitrageurs in the country, typically with $500m or more in annual revenue can justify the expense.

    It is important to stop this senseless race of shaving off another nano-second; there is no increased productivity once the refresh lag drops below human reaction time, roughly 200ms, a milestone that was passed in 2004. On the contrary, there is actually tremendous downside in this crazy speed race, as witnessed two years ago in the “flash crash”. Actually, there are plenty of simple and natural remedies, such as charging for rapid cancellations of orders on the book for less time than human can react to, but the US equity market is “self-regulated”, i.e. the SEC does not set the detailed rules; the exchanges do. The root cause to all this silliness can therefore be traced to the fragmentation of the equity (i.e. stocks) markets in the US. As each of the four major exchanges compete for order flows, they become beholden to the largest traders, and nobody can do more volumes than high-frequency firms.

    Finally, just two weeks, the NASDAQ (in my opinion, the most professional and sensible exchange in the US) decided to take a small step in the right direction. A small fee will be charged on frivolous order placements/cancellations. So, maybe speed will lose its allure and no one will be funding the neutrino idea…

  12. Chris Oakley says:

    In 1994, when I worked on the Swaps Desk at UBS in London we planned to set up a neutrino communication link with with Sydney office. Originally protons were to be accelerated to 99.99% of the speed of light and smashed into a Nazi gold target, but to cut costs, it was decided to use Argentinian government bonds instead of protons. Unfortunately, this did not work as the bonds would default before reaching the target.

  13. Anon says:

    Well, if neutrinos are FTL, that could provide a whole new mechanism of shorting a stock. :)

  14. Ted Baxter says:

    Dare I ask what a Nazi gold target is?

  15. Chris Oakley says:

    Ted,

    I don’t wish to labour this but UBS = Union Bank of Switzerland. A lot of gold is stored in vaults in Zurich by Swiss banks including, allegedly, that originally stolen by the Nazis. I doubt whether, in reality, they would allow it to be used in a scientific experiment, especially if it ends up being radioactive. The chiefs would often play lip service to globalization but the reality was that the various centres (London, Zurich, New York, Tokyo and Sydney) would do their own thing. The only time that fast communication would have made a difference for our group was when Zurich & London had a video conference with New York – we actually had to fly to Zurich to join it and as well as a huge delay, the picture quality was poor (this was 1994/5).

    Going back to the original topic, I doubt that program traders would benefit from a few tens of milliseconds speed advantage, and if they did, it is questionable whether such trades should be allowed as it would almost certainly be because someone was trying to game the system.

  16. Owen Patterson says:

    Let Wall Street and its global coven think they need neutrino communications. They will pay for all of it and then the science community can take it off their hands and get some real work done.

  17. AnonToo says:

    M.Wang, although your comment is well-informed (which I very much appreciate given all the misinformation out there on finance), and I agree with much of what you say, I’ll have to disagree with this:

    It is important to stop this senseless race of shaving off another nano-second; there is no increased productivity once the refresh lag drops below human reaction time, roughly 200ms, a milestone that was passed in 2004. On the contrary, there is actually tremendous downside in this crazy speed race, as witnessed two years ago in the “flash crash”

    First, the flash crash was a confluence of many causes, including an extremely large price-insensitive sell order, and I’ve never seen a report (speculation, sure, but nothing formal) that indicated HFTs were to blame. Of course HFT behavior is important, since it’s a huge fraction of the market. But I don’t know of a theory that says HFT caused the crash. If anything, it was HFT pulling out that caused the liquidity to disappear. In fact, the SEC was even considering adding rules that HFT can’t “turn off” like they did!

    Second, the flash crash was not a “tremendous downside”. It was something like 20 minutes. It had no consequence to most investors, and errorneous trades were busted. A few people learned that stop-loss orders with no limit are a bad idea (which was true before the flash crash, too), but overall there were few stories of real pain and there was no domino effect.

    Third, I think you are vastly underestimating the benefits of HFT. There most certainly are benefits to sub-200ms trading. The liquidity in stocks with HFT participants is tremendous–as measured by bid-ask spread and cost of executing large orders–compared to stocks without HFT participants or stocks 30 years ago. We’ve replaced a large number of slow humans taking huge bid-ask spreads with a smaller number of very fast computers fighting for a few pennies. Needless to say, the latter is much more efficient (in the sense of market efficiency AND use of human talent) and is a better deal for investors.

    Sure, it’s silly people can make a lot of money reducing their network ping. But it’s diminishing returns, so it won’t last forever. Over the next 20 years (uneducated guess) things will stabilize as traders pull out the last bits of arbitrage that are still left.

    We might as well accept the silliness for now, and be happy there are people investing in communication technology, no?

  18. Bugsy says:

    Z:
    Interesting about the submarine communications. So, you need 1o seconds to transmit the one bit “fire missiles” and a bit more to transmit the more complicated “Oops- scratch that!!!”

Comments are closed.