Nanosecond trading could make markets go haywire
February 24, 2012
Red line represents the frequency of sub-650 millisecond flash crashes, and blue the frequency of flash spikes, between January 2006 and February 2011. The black line is the S&P 500 index. (Credit: Johnson et al./arXiv)
A new study says “flash crash” events happen routinely, at speeds so fast they don’t register on regular market records, with potentially troubling consequences for market stability.
The analysis involved five years of stock market trading data gathered between 2006 and 2011 and sorted in fine-grained, millisecond-by-millisecond detail.
Below the 950-millisecond level, where computerized trading occurs so quickly that human traders can’t even react, no fewer than 18,520 crashes and spikes occurred. The study’s authors call those events “financial black swans.”
The programs are designed to trade enormous volumes of stocks, bonds and other financial instruments at superfast speeds, taking advantage of second-to-second fractional price shifts and market trends.
It’s now estimated that high-frequency computer trading accounts for 70 percent of all equity trades. While some activity does occur at speeds with which humans can interact, much of it falls beyond the limits of human response time.
Ref.: Neil Johnson et al., Financial black swans driven by ultrafast machine ecology, arXiv, 7 February 2012.
[ Wired ]
Related Site Content:
- Algorithmic stock trading rapidly replacing humans | September 9, 2011
- Planets could orbit singularities inside black holes | November 10, 2011
- The $300m cable that will save traders milliseconds | September 14, 2011
- If a primordial black hole hits the sun… | June 6, 2011
- How seasonal moods affect investments | October 12, 2011
Comments (5)
by eldras
Rick i dont know expert trading but any move to ditch the casino and move to value investing would be good. While futures have a part to play clever arbitrage is not related to individual companies
by Rick B
Thisnis an especially dangerous and completely unregulated phenomena. We work as day trader and can see that humans no longer can compete withthe algorithmic systems and we get hammerd on prices. The system also works to block “real” traders from access by overwhelming the market with floods of bid asks. Also, the Market makers are no algorithmic as well and many large funded institutions are working in concert to force stock prices. It was proposed and shot down immediately by politicians to insert a 50 microsecond delay on every transaction. I think 200 milliseconds would be better which would permit humans to once again be able to compete. Not everyone is a large fund and this is an unfair advantage to what is supposed to be a free market. Nontheless these systems are not performing well and are based on a sttistical average of gains of 1 or 2 percent a day fom millions of transactions. Many systems can’t even do that and perform far below human levels of performance so thus at least up to now not worth the billions being invested. We badly need scontrols to prevent these systems working in concert to destroy the financial markets.
by eldras
The hope is that markets move to perfection and that biology is a better model (Rothschild).
Humans are part of the universe as therefore moving by the laws of the universe and therefore their actions and tools are moving by eventually predictable laws.
in order to accurately predict you have to be a higher system of intelligence ,or predictive modelling. Superintelligences that must emerge as weak A.I.s have already begun being built eg in expert systems, should be able to model our present markets behaviour, and these are expected to begin appearing about 2 years after a simulation of a human neocortex in 2018, as hierarchy intelligence classes are automatically built up and networked.
I agree with Vinge and de Garis these are inevitable consequences of human technology and dont need to be specifically designed.
by he who is
Except AIs don’t (yet) beat out the most intelligent humans out there. Only in narrow fields can they even achieve moderate to high proficiency, and only with great fine-tuning. Even then, we have these nanosecond scale crashes, which are completely meaningless fluctuations of a virtual monetary system that could quickly lead to an unprecedented crash that has drastic and tangible effects.
The consequences of advances are not being considered before we begin to take advantage of such advances… And this is creating huge potential for disaster!
by peter g
the trading robots are taking over