Artificial Intelligence (AI), again one of those terms that everybody talks about yet very few actually have a clue. Loosely being used interchangeably with Machine Learning (ML) and seemingly has much to do with data analytics in general, most have to claim these days that it is being utilised somehow or else risk being deemed out of touch in the investment world.
While AI is certainly no spring chicken and said to be in existence since the 50s if not even pre-war, our recent advancement in data analytics from computing power to storage capacity, as well as the speed of transmission, have all leapfrogged to such a level never seen before in the history of mankind. On top of that, the amount of data created and collected in just the recent 2-3 years was also said to be comparable to the sum of ALL those that were being created beforehand. As one teenage AI protege puts it, it is not that the younger generation is any smarter, it is just that the fire power in their hands was never available before. Yes, AI is thus here to stay and it can only get better, and almost certainly in a even faster pace. Any pro in the investment space probably has no choice but to brush up one’s vocabulary on it before it is too late.
But no matter how shinny your new toy is, getting the job done is the only thing that matters in the end. In the investment world where returns outweigh everything, this is indeed the holy grail. It has certainly been already a few years that everyone and his dog claim that they have some sort of ‘proprietary’ AI-enhanced models that will help them stand out from the crowd. The reality, is something that consistently deliver is very hard to come by. When even the biggest industry giants privately admitted that AI probably contributed less than 10% to their returns these days, any claim of AI revolutionising the investment scene is still too early to call. What most are currently practising, from what we can conclude so far, is certainly the much heavier usage of data, big and small.
Machines are indeed being ‘trained’ (note: not learning by themselves just yet) extensively these days, in and out sample, to try to come out with some sustainable alpha generators. As most seasoned in the field will admit, this is however no guarantee of any meaningful results nor breakthroughs. A typical AI ‘plateau’ of fruitless attempts in the dark alley can easily sink you back a few years, and they certainly won’t come cheap. Even if some kind of alpha is indeed being explored, it will almost certainly be soon arb-ed away by an influx of capital. In today’s world of ever growing computing power, it is basically a race against time, and of course the depth of one’s war chest.
The good news: seemingly none of our jobs are to be taken away by AI anytime soon when it comes to alpha generation. It is still going to be human instincts, combining with experience, that is going to make the difference. Of course, a bit of data analytics on top that comes with a sexy name won’t hurt!
