Algos and mindlessness


The much predicted impact of high frequency trading is starting to be felt on the Australian stock market. A program on Background Briefing entitled “Attack of the Algorithms” put the level of algorithmic trading at 40% of the turnover. It is on the way to the 70% that is the level on the NYSE. There were many concerns expressed about high frequency trading (HFT). Plus there was an unconvincing apologist from ASIC, surprise, surprise. ASIC can be counted on to have its eyes closed when it matters. Yet what I found most compelling is how metaphors divert analysts away from the real problem.

The metaphor, of course, is one of fluid, or water. HFT provides “liquidity”. What matters is “deal flow”, one algo trader said. It creates the illusion that financial behaviour can be analysed like tides or waves, a sort of tangible phenomena that can be observed from a distance. In fact, it is a set of rules and the question is who gets to set those rules. The traders or the government and the regulators acting on behalf of the government?

The usual arguments were on display. “Oh well, it’s all inevitable, because, as with the explosion of derivatives markets, traders are much smarter than regulators, so no matter what regulators do traders find a way to defeat it.” This assumes impotence of the government, which is only the case if the government decides not to govern. It would be possible to ban algorithmic trading, or to go to some lengths to limit it, like email spam. As far as the efficiency of the stock market is concerned, nothing would be lost except volatility.


I have already talked at length about the myth of financial de-regulation, which is nonsense when we consider that money IS rules. This argument: “Oh, traders are far too clever to be regulated” is just the other side of that coin. It is simply saying that the cleverest get to set the rules, simply by virtue of their cleverness. One would hardly bowl up a similar argument to justify theft, for instance. “Oh, he was a clever thief, so I guess it was fine that he stole all our money.”But in the finance sector, such nonsense is accepted.

Indeed, algorithmic trading is an extreme exemplification of the point that money IS rules. Algorithms are only possible if there is a rule based system on which the equations are based; mathematics requires basic laws in order to exist. Is it really impossible for the rule setter, the regulator, to control that, given that the algorithms’ very existence is dependent on the conditions the regulator creates?This posture of helplessness is a deep, continuous abrogation of responsibility by government which is endemic across the financial markets.

But of course on we go, helpless in the face of trading ingenuity. Which is creating a very interesting tension between man and machine that I expect will shape much of what happens in financial markets over the coming period. Because we have failed to control the techniques for gaming the system, we are entering a kind of cyborg period when the machines will start to control us. Flash Crashes are probably only one of the problems. Humans will start to exit the market. The Background Briefing episode interviewed a number of smaller traders who basically said they were getting out because they can’t compete with the machines. It is only likely to get worse.


Allowing such tampering with equity markets is deeply irresponsible because equity capital, which is based on a higher order of trust that debt capital, is critically important to avoiding or getting out of crises. Equity, as was seen in the $100 billion raised by listed Australian companies in the wake of the GFC, is critical helping firms buffer themselves against crises. And when there is a crisis, equity can reprice without imperilling the system, which is not the case with debt, especially bank lending. Equity is the balm of the system, especially English-speaking financial systems, so it is no small thing that it is being gamed.

The philosopher of science Stanley Jaki made this comment about artificial intelligence:

Herein lies the worst fallacy of the whole modern discussion about computers as artificial intelligence. Machines do not add, they do not calculate, they do not integrate any more than a gutter does not add or integrate by being the channel for millions of raindrops. In an electronic computer not raindrops but electronic impulses are channeled along strictly predetermined routes. In the process no addition is performed. It takes a mind, always a mind, to abstract meaning from each step through which the machine is directed by its specific man-built mechanism.


It takes a mind, always a mind, to calculate the worth of a company or an investment. It takes a mind to provide the much lauded “price signals” about a stock. But apparently you no longer require a mind if you are a regulator or government washing your hands as the algorithms are allowed to take over. Perhaps we can automate government as well.

In the final analysis, of course, the machines can’t rule. We created them. But we seem determined to learn that lesson in the worst possible way.