Investment “Systems”

I recently commented as follows to a student just starting out in finance:

Jonathan

If you really are a student and just starting out down this road let me offer you a little advice. Although I may of course just be plain wrong. Many are now saying that machine learning and AI are the name of the game, not old fashioned indicators such as these.

You need to look at the arguments for and against. I’m not at all sure at this stage ML is going to lead very far in investment terms based on the most fundamental and top down view.

Markets are not a “fixed system” like the areas in which ML is seeing most success. E.G. Go , chess and driverless cars. We do not know the rules of investment. We do not know if there are any long term rules or whether the rules simply keep changing with the years. Even if we did know all the rules (variables) it is doubtful we have the data or computing power to compute the relevant market forecasts. And that is to leave aside the question of whether markets are deterministic or random.

So frankly, these old fashioned trend following methods you are looking at may be as good as you can get at present. You need a trend to make money.

There are ways to make big money out of markets I believe but I do not think you will find them on this forum. Inside information, dealing ahead of clients, the bid offered spread at high frequency and so forth.

For long term investment, I rather wonder whether people on this forum are not simply chasing their tails, including Quantopian. Long term investment is simply about following the trend of economic growth. I am not convinced there is much more to it that that. It is about diversification over world economies and currencies.

It may well be that attempts to improve on simple diversification and buy and hold are not terribly useful. But it is fun to try.