In the past, the stock market was not exactly predictable, but it had a certain personality. It would have highs and lows, it would move in cycles and prices would eventually return to something resembling the underlying value of the company whose shares are traded. Because the market is based on human trading, human emotions would drive market movements. Fear and greed would push the market to extremes, and once the weak traders had left the market, it would revert to the mean.
I have a friend who has a PhD in math. He told me about someone who watched the market for momentum (emotional swings) and mean reversion. For a while, this worked, much like my weekly market outlooks that consider momentum and value. But it doesn’t work anymore.
Here’s another example of something that doesn’t work anymore. I decided to look at the stock market the way a farmer might look at weather 100 years ago, to find patterns that would give some predictability. For example, spring is usually rainy and winter is usually cold. In the same way, June and September are usually pretty rough in the stock market, but December is usually profitable. While that worked for the period from 1979 to 2009 (or so), it doesn’t appear to hold any longer. As an example, the stock market rises on average 60% of the time. In June, it rises less than 50% of the time, and in December it rises over 80% of the time. But recently, and most likely in future, the market rises 60% of the time in June and in December and every other month.
My friend suggested two reasons for this fundamental change in the markets. There are probably more. First is high frequency trading. There are traders who have computers that are fast enough, and access to the market, that is fast enough, that they can front-run trades to make a fraction of a cent on every share traded. Front-running is illegal, but markets (eg. NYSE, TSX) allow this with the excuse that it creates liquidity. What it does is create risk-free profits for some traders at the expense of others. The book Flash Boys describes some of this.
The other reason is machine learning, which he called artificial intelligence. Computers now have enough processing power to find patterns, the way people do. Having said that, they see patterns differently, and patterns that we might not see at all. We now have computers finding and exploiting patterns in the market, faster than people can. If you want to see evidence of this, look at the recent performance of hedge funds. Whereas they used to be able to use their experience, research, insight and computing power as advantages to make money, now they are hardly able to beat the machine-learning-enhanced market. Here’s a glimpse into what it takes to find patterns in the stock market: John Simons interview. Sell-side trading desks are experiencing similar challenges.
If you can’t beat the market, what can you do? Buy the market. It will rise 60% of the time. It will create value over the long term. Buy and hold on tight. Diversify, so that you don’t have to worry about the operational risks of an individual business. Think long-term. The shorter and faster computer trading becomes, the greater advantage we have in investing for the long-term. In fact, I believe we have little alternative.