For finance geeks and stock market punters, here is an article about the growing use of computer programmes to trade the equity, bond and other markets. Even as early as 1987, when equities fell dramatically – was it really nearly 20 year ago? – I vaguely recall reference to ‘programme-trading’, a process whereby orders to buy or sell a bunch of stocks was automated. Banks and hedge funds now use what are called algorithmic trading systems, which, in plain English, make use of recognisable patterns of behaviour that can be expressed mathematically in order to give out ‘buy’ or ‘sell’ signals in a market, spot trends, etc.
The usual worriers, not all of them anti-market people, may fret that all this mathematical wizardry, aided by the powers of modern computing, will make markets dangerously volatile, but as Iain Dey’s Telegraph article suggests, this does not appear to be the case. In recent years, in fact, global equity and bond markets have been pretty calm, although punctuated by the occasional sharp selloff, as happened in late February and early March. The last really big blowup was when Russia defaulted on its sovereign debt in 1998, triggering the meltdown of Long Term Capital Management, a hedge fund. When last year the fund Amranth nearly collapsed in the natural gas market, it hardly caused a wider ripple.
In fact, contrary to what the Will Huttons of this world might have us believe, the growing use of financial derivatives to offload risks seems to be making markets more, not less, able to deal with risk and ultimately, makes the whole financial system safer. That is not, of course to say that all is well. It is not. In Britain, a profligate government could yet put the market into a spin if the inflation problem gets worse (UK retail price inflation is nearly at 5%). House prices could, if interest rates rise as expected, take a nasty fallback. So there are gremlins in the systems. But the blame, as usual, should be pinned on the real culprits, and not computers or strange-sounding things like collateralised debt obligations.
Of course, this also explains why some of the best science graduates and post-graduates now work in the City, rather than making space rockets. Money talks.
(I have corrected the spelling of Iain Dey).
The Crash of ’87 and the suspicion that program trading contributed to it persuaded the New York Stock Exchange to institute the “circuit breaker”. That is a rule by which if the market falls by a given amount, the exchange suspends trading for two or four hours or the rest of the day. This doesn’t apply, of course, to foreign markets or other U.S. exchanges.
Anyone who could consistently predict short-term swings in any investment market would have the key to the mint but no one has ever done it long enough to be better than chance. The computing power of today’s networked computers is vast, the programmers as ingenious as ever, but the computing problem is still too complicated for them. Maybe, someday, someone will invent a trading program which consistently works but even that is likely to be self-defeating once knowledge of it spreads since when the computer program which models the investment market begins to affect that market, Heisenberg’s uncertainty principle takes hold.
“Heisenberg’s uncertainty principle takes hold.”
I suspect it is a good deal more uncertain than that. Not only would you get chaos, in the technical sense of extreme sensitivity to initial conditions, with a very large set of initial conditions, but other trading machines would adding functional dimensions rather unpredictably even if non-machine traders could be effectively modeled.
What would happen if all traders (AI or the red-braces type) were equally as good. What would happen (this is stronger) if they all had perfect information?
I appreciate the former is highly unlikely, and the former practically impossible. I’m just curious in a tecnhical sense.
He is called Iain Dey, by the way.
(I did a double-take seeing my name in the posting).
Disciples of Nicholas Nasim Taleb will know that absence of turbulence does not mean that turbulence will not recur; it is more likely that stability is the precondition for the next outbreak or turbulence
Chaos theory?
Damn!
So my three engineering degrees are all useless then.
Not necessarily. The history of finance shows you can get very well paid for appearing to know what you are talking about and/or being able to manage very sophisticated calculations about the consequences of some abstruse hypotheses.
Believe me, the universe really is causal.
Sometimes we understand how.
Sometimes we don’t.
But talk of ‘chaos’ is bitter surrender to a futile outlook.
The fact that the universe isn’t ‘chaotic’ is what permits believers in chaos the time to contemplate their own resignation.
Of course, it would be much more convenient if the universe was Subjective; then we Engineers would be all that was left after the chaos-believers spontaneously dissociated.
When the credit money supply is expanded (for example when Ben Strong, Governor of the New York Federal Reserve Bank, agreed to expand the credit money supply of the Dollar in the late 1920’s to support the exchange rate of the British Pound) the new money must go somewhere.
The real estate market and the stock exchange are classic places where the new money goes.
It is not really the fault of the people who play with the credit-money bubble (by whatever clever schemes they play with it) that the bubble eventually pops.
Although many people often blame “city types” for things they never really started.
Contrary to the popular myth, people in the financial services industry normally get out of these things well before they go pop. I do not blame them for this – making a profit from a credit-money bubble that one did not create is not wrong, and getting out before it goes pop is only sensible.
But talk of ‘chaos’ is bitter surrender to a futile outlook.
No it isn’t. It is a factual description, not a moral judgment.
Chaotic systems are not always indescribable or unmanageable even if their exact configurations cannot be predicted.
If chaos was the only problem, it would be easy. Chaotic systems are predictable in the short term, but only the possible range of behaviours are predictable in the longer term. The stock market, conversely, is very unpredictable in the short term but has turned out to be quite predictable in the long term (i.e. it goes up exponentially).
The problem with the stock market is that everybody is applying the most sophisticated algorithms money can buy to spot any faint hint of a pattern and exploit it, and the process of exploitation naturally removes it. (If you can predict the price is going to go up, that makes it that much more valuable now, which means if people know about it then the price will be already there.) This has the effect of explicitly removing any predictable features it might have, leaving precisely all the stuff we can’t predict yet.
There is no reason to think markets will be any more volatile with computers running them, and some limited reasons for thinking it will be less. But there will never be any way to entirely remove the unknown (or at least, not until we invent time travel).
If one wishes to invest in the shares of a company one should examine this company carefully (not just look at the books, they could be a tissue of lies – as my father found to his great cost with “Slater-Walker” many years ago) and then make a judgement about both how good that company is and what the general economic conditions will be over time.
If one wishes to invest in a commodity one must study and make a judgement over what the supply and demand for that commodity will be over the period of time one wishes to invest.
As for clever maths – no polite comments can be made.