High Frequency Trading and the return of the pit

July 25, 2009

Following up here on yet another great post by Martha. High Frequency Traders (HFTs) employ trading strategies that are dependent on very quick conditional buy and sell orders. These orders are aimed, for example, at exposing the typically well-hidden order limits and size of orders of other traders. The speed of order production is based, in turn, having as few mediators in the communication network between the traders’ computer – generating the trading orders – and the exchanges’ computer that executes them. In a world where every millisecond counts, event at light-speed speeds, physical distance counts. In fact, we should say ‘physical distance’ counts AGAIN. Yes, physical distance, which used to be an all-important factor in the face-to-face trading pits, is now making a comeback in the form of vicinity from the exchanges’ servers.

This, of course, brings back the issues about the politics embedded in market spatiality. Kate Zaloom shows how important were the top steps of the trading pit in the Chicago futures exchanges, and, consequently, how the political (and sometimes physical!) struggles for these coveted locations. Can we expect similar fights to rage over positions in the electronic communication network that transmits trading orders? Judging from the evolving competition for ‘electronic proximity’ the answer is positive. For example, Wall Street & Technology reports that in 2007 about a 100 trading firms relocated their serves into Nasdaq’s trading headquarters, just to be close to the executing servers and to shave off about 7-35 milliseconds from the communication time (depending on previous location of trading servers). Similar trend is evident in NYSE and the Chicago futures markets.

What does trend teach about the techno-social nature of markets? From an actor-network perspective, this is yet another evidence that markets, just like any other social institution, do not exist in a baboon society. That is, materiality, in general, and tools and devices, in particular, are delineating and indeed, shaping social interaction. There is at least one more insight to be gleaned here, though. It is true that physical space is reintroduced to markets through the relocation that comes along with HFT, but something much more important is being introduced to market with it: the rich, nuanced information that used to be the lifeblood of face-to-face trading and the infrastructure of liquidity supply is now communicated electronically. Techniques such as ‘pinging’ and ‘fleeting orders’, both based on algorithmic, high frequency trading (among many other techniques) are the electronic equivalents of a specialist muttering a clue, under his breath, to a nearby floor broker. It used to be the commonly held belief that electronic markets offer a distinct dichotomy between placing an order and not placing it; a distinction that could not have existed in face-to-face market were clues, partial disclosure and innuendos were an immanent part of trading. In fact, that claims about electronic markets’ improved transparency were based largely on this techno-social characteristic. However, HFTs seemed to have blurred this distinction and even question whether it actually ever existed. In a paraphrase on Latour, we can ask: have we ever been electronic?

14 Responses to “High Frequency Trading and the return of the pit”

  1. Fabian Says:

    Polemics of scooping, of last-second trades and of speed asymmetry have been at the center of market design controversies for long. This paper here (also available here) is about the case of call auctions as price pacifiers (in the 1990s, Paris Bourse).

  2. yuvalmillo Says:

    Yes! Absolutely! The issues related to the inherently imperfect translation (or ‘folding’) of face-to-face markets to electronic ones are not new and only reappear in a different disguise every so often. The paper you cite here is a gem, which is (unjustifiably) little known, because it is really pioneering in its approach to the topic. Daniel and I, naturally, refer to it.

  3. danielbeunza Says:

    Very nice last sentence in your post. And I agree it is indeed fascinating that an early electronic peek at the orders coming up resembles so closely the stepped surface of a physical trading pit. And it’s even more interesting to note that asymmetric information is not all that counts. Ambiguity also has a place. Flashing and fleeting are the new winks and nods.

    However, is the analogy too beautiful to be true? What I heard from the New York representative of BATS is that locating close to the exchange/server story cannot be true… because exchanges don’t keep servers in their buildings. Would be interesting to really establish what’s going on here.

  4. danielbeunza Says:

    Fabian, what is a price pacifier?

    Your paper, which I read in its multiple incarnations, evolved from a performativity approach to an argument about signs.

    The opening and close auction are the only parts of the NYSE trading day that have not been automated. Do you have a theory for why that might be?

    • Fabian Says:

      The North-American call auction debate in the 1990s refers to the idea of a NMS, the attempts at automating the NYSE, sunshine trading, the AZX and the struggle against volatility and manipulation at close (hence the idea of cooling down, pacifying prices). In the Parisian implementation, the idea of countering high-frequency gaming at close was explicitly at stake.

      (The paper uses that for an argument on the representativeness of prices.)

      • danielbeunza Says:

        Very interesting. The issue, I guess, is what do “non-representative” prices mean in terms of the current political debate over these issues?

        One possibility is, more volatile. But in my conversations with one of the head high-frequency strategists at Goldman, he pointed out that although very short term volatility has increased, if you reduce the frequency at which you measure volatility –and he argues that the relevant volatility is not the minute-by-minute one– then volatility has stayed the same.

        Your view?

      • yuvalmillo Says:

        I think that we have to separate, albeit related, topics here. One is the issue of quote manipulation through last-minute-before-closing trades and another one is the story of flash orders. The manipulative trades were intended to change the end of day closing – they were not initiated for their transaction value. In contrast, flash orders are meant to be accessed by counteracting trades and the entire techno-social setting is designed for bona fide trades.
        Now, having said that, I think that are interesting connection points between the phenomena. For example, there is what I would call – the ambiguous intentionality (I’m sure lawyers have a better word for it) – it would be relatively difficult to distinguish between a bona fide last-minute trade and a manipulative one and any distinction, if one could be made, would be done on the basis of relative quantities on not a positive qualitative basis. Of course, distinctions were made and way closing prices were calculated changed, as Fabian’s work documents. Similar situation exists in the flash orders story. The exchanges that offer flash orders and sell access to these orders know that what this does not comply with the intention behind the one-second provision of Reg NMS. Yet, there exists intentional ambiguity about the things that can and cannot be done during the one-second provision. To take a gamble: in this case, just like in the case of the closing prices, we will see tightening of the distinction rules.

      • Fabian Says:

        “if you reduce the frequency at which you measure volatility –and he argues that the relevant volatility is not the minute-by-minute one– then volatility has stayed the same”

        The guy is perhaps referring to “high-frequency finance”, i.e. playing with the clock (e.g. changing calendar time for order-book time) while analyzing the distribution. See: http://www.jstor.org/pss/222489

  5. marthapoon Says:

    Columbia Journalism Review “reviews” coverage on HFT:

    http://www.cjr.org/the_audit/the_times_takes_on_highfrequen.php


  6. With current technology, we could run call markets once per second, removing most of the problems caused by high-frequency trading.

    http://mblog.lib.umich.edu/strategic/archives/2009/07/countering_high.html

    With respect to the point about volatility only being increased at the finer time scales, this still represents a significant efficiency loss and a transfer of surplus from regular investors to rent-seeking traders.

  7. marthapoon Says:

    Goldman Sacks clarifies it’s position on HFT:

    http://www.zerohedge.com/article/goldman-sachs-
    clarifies-its-high-frequency-trading-practices

  8. yuvalmillo Says:

    Martha – thanks a lot for this link. You can almost feel the panic in the air (“Sir, I swear I was not using Flash orders! He did, sir”)… But, more seriously, HFT is likely to play a major role in the political process where dark pools are constructed as the current villain of the financial system (in a way, similar to the role that stock options played in the Congressional debates that preceded the enactment of the SEC in the early 1930s). See, for example, http://www.tradersmagazine.com/news/sec-flash-orders-ban-schumer-104174-1.html?zkPrintable=true


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: