The summer holidays are always a good time to catch up with some reading, and I brought ‘The Fear Index’ by Robert Harris with me to Southern France. I’m sure many readers will already have read it, as it’s been around for a couple of years now, but it was new to me, and I found its central thesis pretty relevant to current markets.
Algorithmic Hedge Fund
The plot concerns a mathematically-gifted but close-to-aspergers computer whizz who has given up his scientific role at CERN to become the co-founder of and major force behind an algorithmic hedge-fund. The book is well-plotted and Harris has clearly spent the time and effort to learn about financial markets with the result that it reads plausibly, even to those with their own background in trading, which is not always the case with financial fiction.
Artificial Intelligence
There are various sub-plots to drive the action along, but the basic premise of the book is that Alex Hoffmann (the geek turned fund manager) has developed an algo trading system to run the fund. The particular twist is that Hoffmann’s speciality is artificial intelligence, and his system is self-teaching; in other words, it learns as it develops, rather than simply running the same algorithm constantly. Early on in the book, for example, there is a plane crash and it turns out that the fund has scored hugely through being short the stock of the airline carrier involved. The scary bit is that Harris plants in the reader’s mind the suspicion that VIXAL (the name of the programme) has been somehow able to engineer the accident in the first place, in order to profit from it. Other strange events occur, all pointing to the inevitable conclusion that the programme is running out of control.
The Flash Crash
This is not a book review, and I don’t want to give the plot away too much to anyone who hasn’t read the book, but the finale builds towards the ‘Flash Crash’ of May 2010. The Fear Index of the title is, of course, the Chicago VIX, the volatility index, which plays its significant part in VIXAL’s trading strategies.
‘Fat Fingers’
Now, I’m not sufficiently knowledgeable of technology to know how far down the road to the independent artificial intelligence that Harris describes we have gone, but nevertheless the incidence of seemingly random uncontrolled bouts of trading – basically, so-called fat fingers trades – would appear to be rising, the latest one being in Shanghai just a couple of weeks or so ago. I wouldn’t go as far as Harris and suggest that perhaps these are deliberately engineered by smart computer programmes to achieve their own ends, but it’s pretty clear that it is bugs in algo-trading systems rather than human traders genuinely miscuing their keyboards with clumsy fingers which cause the problems.
Exchanges and regulators globally watch this kind of event carefully, and are on the lookout for strange trading patterns. They do their best, with circuit-breakers and the like to try and isolate any problems and maintain the overall integrity of the markets, but I just wonder if the way sanctions are normally applied is as effective as it might be. When a fat-finger trade occurs, and we see a market bent seriously out of shape, the regular approach is that the company involved is slapped with a regulatory fine and the suspect trades are cancelled. At first, that seems equitable. But then, on reflection, perhaps less so.
Are the Sanctions Equitable?
Consider this example: in a very thin LME market, a computer trade sells copper down a few hundred dollars through a limited volume of bids. Within a very short space of time, as we know from experience will happen, buyers – human or electronic – will observe the oversold price and put buying into the market which will then recover. The broker involved will then call the LME, explain that they had a system failure that resulted in their selling pushing the price down creating trades at a substantial discount to where the market ‘really’ was. The LME sends out a notice cancelling the rogue trades and at some point there is a regulatory finding of what actually happened and who was to blame. Then there will probably be some sanction. But there are two sides to every trade, and it feels slightly uncomfortable that the buyers in that example, who had, let’s say, resting scale-down orders in the market have not in fact made the trades that they thought they had. OK, they were well out of the market range, but suppose the first execution had triggered other trades elsewhere, as part of an overall strategy. One leg of that strategy has arbitrarily been taken away.
Consequences and Responsibility
There are consequences to actions, and with the increase in fat finger trades, it may be that the time has come to strengthen the sanctions and question the practice of cancelling trades created by an error. There will always be mistakes, and clearly nobody deliberately writes a programme with a glitch in it, but perhaps the knowledge that you would have to stand behind your algo’s errors (as well as possibly facing fines) would add a little piquancy to the testing and proving of systems. Responsibility for actions is an important part of the reality of life and one person’s (or system’s) error has consequences for other people’s behaviour – shouldn’t that perhaps be recognised? It’s a tough choice: neither solution seems comfortable, but on balance, wouldn’t it be preferable to go with enforcing responsibility?
Of course, it would probably not have changed anything in Robert Harris’ book, but that’s part of the luxury of fiction.
