Quant-ifying the HFT effect in stock movements

Discussion in 'Professional Trading' started by ASusilovic, Jul 13, 2010.

  1. Here’s something for critics of high-frequency trading (HFT) — and lovers of Markov/Gauss/Mandelbrot minutiae — to pull out for their next dinner party conversation:



    It’s a quantitative analysis of 14 NYSE- and Nasdaq-traded stocks between the start of 2002 and the end of May 2009 — including HFT favourite GE . The idea is to look at these shares to determine whether HFT has had an impact on stock movement correlations.

    And, in a word, it has. At least, according to the author.

    In particular, this paper will demonstrate, that since 2005 there has been a marked changed in the correlation structure of stock trading dynamics where amongst many stocks, there has been a measurable departure from the typical H = 0.5 regime and that stronger self-similarity has been steadily increasing over the same period in the time that HFT has become the largest source of market volume.

    Without going into a mass wonk-out, the H here stands for Hurst Exponent — a numerical estimate used in fractal geometry and also known as the “index of dependence.” It’s basically a way to measure the repetition or predictability of a time series — in this case stock movements.

    H values range from 0 to 1, with a value closer to 1 indicating more predictability, or more “self-similarity.” Think of fractals — a pattern in which the overall pattern is repeated in miniature within that pattern — and the way in which they echo themselves, or parts of themselves.

    And, according to previous research of short-term intraday equity trading data cited in the paper, that H has tended to be roughly be equal to 0.5 — indicating a Brownian time series (stay with us) or basically a random walk. In other words, there is little or no correlation between one value and a future value.

    However in recent years that value has somewhat changed. As the paper puts it:

    For the NYSE stocks, the Hurst exponents increase from 2002 onward but by late 2005 barely break the average of H = 0.55. Therefore, during this time (and before), short term trading fluctuations do not appreciably depart from an approximation of Gaussian white noise. However, once Reg NMS is implemented the structure of the trading noise begins to change rapidly increasing to 0.6 and beyond in a couple of years. This is a new behavior in the high-frequency spectrum of trading data that indicates increasingly correlated trading activity over increasingly shorter timescales over the last several years. Correlations previously only seen across hours or days in trading time series are increasingly showing up in the timescales of seconds or minutes.

    A more complicated picture is shown in the NASDAQ data. The Hurst exponent of NASDAQ stocks started rising much earlier, from 2002 or earlier. By the time Reg NMS was passed, most NASDAQ stocks already had an H which many stocks on the NYSE would not reach until 2009. Some NASDAQ stocks, such as AAPL or CSCO, did have spikes shortly after the new rules went into effect but soon returned to normal behavior.

    In fact, data from Figure 10 shows this spike for AAPL was likely not due to HFT as it was most pronounced amongst larger sized trades. It is probable that since NASDAQ was one of the first exchanges to embrace electronic trading and experience HFT via ECNs, the rules officially unchaining HFT for other exchanges from 2005 was close to a non-event.

    Possibly, because HFT was experienced earlier, there is not as dramatic a rise from the time of Reg NMS in H for NASDAQ. The one marked change from the implementation time of Reg NMS is the emergence of massive share trades at the finalet clear data. seconds of trading days, sometimes varying over many orders of magnitude by day, that required the time series to remove to g

    You can see where this is going.

    After the SEC revised the Regulation National Market System (Reg NMS) and implemented Rule 611 in June 2005, the Hurst Exponent for the seven NYSE-listed stocks moved closer to 1, indicating more predictability. For the Nasdaq stocks it happened earlier, for the reasons outlined above, but it still happened.

    Like the global spread of Starbucks in malls and high streets around the world, HFT appears to be making (some) stock movements look increasingly similar — especially for smaller share trades.

    That’s something that would chime with recent accusations that HFT causes mass-correlation and, ironically, encourages more volatility as opposed to more stability. Think of the ‘flash crash‘ of May 6 — when participants pulled out of the market en masse — an H > 1 (+ 1,000?) event if ever there was one.

    The paper’s conclusion:

    Given the above research results, we can clearly demonstrate that HFT is having an increasingly large impact on the microstructure of equity trading dynamics.

    We can determine this through several main pieces of evidence. First, the Hurst exponent H of traded value in short time scales (15 minutes or less) is increasing over time from its previous Gaussian white noise values of 0.5. Second, this increase becomes most marked, especially in the NYSE stocks, following the implementation of Reg NMS by the SEC which led to the boom in HFT. Finally, H > 0.5 traded value activity is clearly linked with small share trades which are the trades dominated by HFT traffic. In addition, this small share trade activity has grown rapidly as a proportion of all trades.

    The clear transition to HFT influenced trading noise is more easily seen in the NYSE stocks than with the NASDAQ stocks except NWSA. The main exceptions seem to be GENZ and GILD in the NASDAQ which are less widely traded stocks. There are values of H consistently above 0.5 but not to the magnitude of the other stocks. The electronic nature of the NASDAQ market and its earlier adoption of HFT likely has made the higher H values not as recent a development as in the NYSE, but a development nevertheless.

    Given the relative burstiness of signals with H > 0.5 we can also determine that volatility in trading patterns is no longer due to just adverse events but is becoming an increasingly intrinsic part of trading activity. Like internet traffic Leland et. al. (1994), if HFT trades are self-similar with H > 0.5, more participants in the market generate more volatility, not more predictable behavior . . .

    Discuss below. Our heads hur(s)t.

  3. In a bull market, HFT is okay. During a bear market, HFT is bad. That is the proper conclusion. :cool:
  4. Well that makes sense. If you want to trade volume X, but have to do it in smaller bites, then one should expect an increase in H since what was once a single monolithic trade has become a big whack of similar smaller trades. H goes up, with no change in actual, meaningful market behavior.