I just had a Ben&Jerry’s with coffee and, as one of the rare creatures left on this planet actually still affected by coffee, I decided to venture off into the night to understand a bit more about what was going on at the stock markets earlier this week. The funny thing is – not much, but a lot. Meaning, that no one quite understands why Wall Street was so jumpy, but the matter of fact is, that it was.
Now, some are throwing around AI trading – that’s when I start paying attention. Because normally I really don’t get excited about jitterish stocks, but AI is one of these modern magic dust dispensaries: With AI everything gets this long-lasting-systemic-effects-angle worth spending a sleepless night over. And even though algorithms and trading aren’t a nouveau thing at all, the world of finance is uniquely qualified to be Ken for Barbie. Or something like that. What I want to say: Super attractive due to it’s inherent basis in numbers as well as clean, accessible and large sets of data.
Back fishing in the ether of the internet I start smirking. A website called ‘The American Thinker’ just published a comment on the stocks taking a dive. “When I think of AI, I think of self-driving cars, which are only as good as the humans programming their cameras and computers.”
Except – Nope. I am always stunned how deeply today’s AI is getting confused with either the 90ies or Terminator. AI at the moment is neither a sentient being able of creative though or social susceptibility nor the sum of what we program it to be. It’s more like a (still very very small) child we are painfully trying to teach not to paint on the walls (as there is an importance difference between this white piece of the world and the white piece of paper on the table). Regarding the trading of stocks that basically means, that deep neural nets are trying to learn how to trade. Or at least to trade better then traders. And looking at the testosterone filled Wall Street trading hall it really doesn’t need a machine learning expert to realize that that’s not too hard to do.
But that’s the short term. Algorithms being simply better at the same thing. What about the long term?
I would say the thing that will really change with AI for trading is long term investment. Humans are notoriously bad at rational assessments of long term risks versus gain. We are too emotional for that. We are too disheartened by seeing the stocks crash.
And many may disagree towards the opposite direction, but I think it’s too easy to assume that with machines everything will just change hands much faster, accelerating the game, trading in a fraction of a second with real-time-analytics rendering human actions either unfit or without any value. To me this seems to only stir the pot of volatility. A scenario like that would not only make absolutely everybody capable of playing almost exactly the same game (exponential volatility), but it would also use the computing power for little (because extremely short-term) gain, while excluding the human factor of risk evaluation to a detriment. Finally, most importantly and with absolute certainty, it would also call upon the regulators to tax those exchanges – as heavily as necessary.
So I guess the perfect symbiosis between the abilities of humans and machines is not just a positive long term scenario but a rather likely one – and it actually rings true for most AI applications also beyond finance.
Yours, Julia | 08 Feb 2018
Update 21 Feb 2018
Netflix’s new documentary ‘Dirty Money’ takes a closer look at exactly this kind of emotional and irrational human investment in its 3rd Episode ‘Drug Shot’ – the shocking case of drug maker Valeant’s glorious rise and inevitable fall.