A key plot device in my book At the Sharpe End is the use of technology in the financial marketplace – specifically, extrapolating trends to predict future movements (I don’t think I’ve given too much plot away in those few words).
So now what do I see in The Australian?
Maybe machines can do better, or at least that is the hope of an increasing number of punters who are using the science of AI for their investment calls.
With artificial intelligence, programmers don’t just set up computers to make decisions in response to certain inputs. They attempt to enable the systems to learn from decisions, and adapt.
Most investors trying the approach are using “machine learning”, a branch of artificial intelligence in which a computer program analyses huge chunks of data and makes predictions about the future.
When do I collect the royalties and from whom?
Tokyo Vice cover
More seriously, this is not the first such real-life event that’s overtaken this book. I was reading the excellent
“Tokyo Vice: An American Reporter on the Police Beat in Japan” (Jake Adelstein) recently (I hadn’t read it before I finished At the Sharpe End for the printers) and I was amazed at how well I’d guessed (or rather extrapolated from my slim knowledge) about things about which Mr. Adelstein has first-hand knowledge.
There was one (fictional) scene involving green tea and coffee involving my protagonist, Kenneth Sharpe and the police, that almost completely mirrored a (true) incident in Tokyo Vice involving Adelstein and the police. And I swear I had no knowledge of the real-life incident before I wrote the fictional account – it was my best guess as to the relationship between a foreigner and the police with regard to liquid refreshment.
It’s not quite AI, but the algorithms are almost complicated enough. The guys a row down from me work on a such a system. They have a lot of problems, and even ones that learn have a hard time understanding drastic situations like the crisis that got us into the current depression. Smarter AIs may be better, but in general such system live in isolation and don’t understand that there are many other systems with similar parameters out there in other firms. When something unexpected comes along it’s easy to turn into a domino effect with the systems playing off another. Even a learning system can have difficultly understanding that a raise or decrease in the market was triggered by it’s own decision to buy/sell a lot of stock that just keeps that snowball rolling.
Banks are trying to build safety nets into their systems to prevent rogue software from loosing the company millions of dollars/yen and keep human traders from making mistakes. It’s still a difficult problem to catch and it’s easy to lose millions due to a software bug.
I’m glad you raised the problem of self-awareness (at least of their own actions) in the trading algorithms (it’s actually a plot factor in the book). On a slightly less sophisticated level than AI is simple program trading, that can also cause positive feedback loops and runaway activity.
What really scares me is the idea of non-linear (chaotic) equations coming into play, (they are probably the basis of most of the market movements anyway). But if all hell broke loose and prices started oscillated wildly with no damping effects, as the result of idiot savant trading systems making all the trade decisions… Not sure. I’m not a mathematician, nor a trader, nor a chaos scientist.