Today's newsletter has just 2 pieces. And they are anything but alike each other. Apart from the fact that one is a short feature and the other one a long cover story, they delve into completely different areas of finance - behavioural and quantitative.
Come to think of it, the fact that one can be talking about cognitive psychology and quantum computing in the same context, that is finance, is what makes it an extremely interesting branch of study. I hope you feel the same way as well!
IN THIS WEEK'S NEWSLETTER
Short Feature - The biases that keep us poor
Cover Story - Using artificial intelligence in portfolio allocation
The idiosyncrasies that keep us poor
Renowned British economist John Maynard Keynes introduced the idea of the ‘Beauty Contest’game likening it to investing in the stock market. Here are the rules of the ‘beauty contest’ game.
People are shown photographs of 100 people and are asked to choose 6 photographs from the lot. The winner is the person whose selection of photos is the most popular across all contestants. In the words of Keynes himself, “prize is awarded to the competitor whose choice most nearly corresponds to the average preference of the competitors.”
So the task, in the minds of the competitors, is not one of choosing the prettiest people. It is not even one of choosing the photos of those faces that other people may find pretty. But it is one of choosing those photos that other people think will be the most popular – This is third degree anticipation.
A lot of stock market investors have reduced the problem of investing to precisely this – Instead of spending their time and energy in finding good businesses, they devote intelligence in unearthing stocks that other people may think are good investment ideas. In some cases, this may work! If we’d picked up Gamestop before the massive rally, we would have made a lot of money. The downside though is that such investors lack conviction to hold these stocks for the long term – which is when money is really made.
Setting that aside, can we really anticipate the average public opinion about stocks? Is this even possible?
Consider a variant of the game played in lab experiments. Say, 100 people are invited to a lab and asked to choose a number between 0 and 100. The winner is the person whose selection is closest to one-half of the average of all numbers submitted. Now, I am sure you figured it out that there is no point submitting a number over 50. And you’d be quite right – if everybody chose 100, the average would be 100 and the winning number would be 50 (one-half of 100).
Now, what if other people are just as intelligent as you are? They will all end up choosing 50. If that happens, then the average of all choices will be 50 and the winning number will be 25 (one half of 50).
Now, what if everybody realised this and chose 25? The winning number will be 12.5! You see where I am going with this…
If you recursively moved down, you will realise that the correct answer to this problem is for everyone to choose 0. The average pick will be 0 and one half of 0, the winning number, is also 0.
In an empirical study with a setting just like this, you’d expect that people will all choose 0. The answer however is far from it! In a lab study, the average pick was 40! And what does that prove? – that there are limitations to rational thinking. The average public opinion does not always have to be rational. And therein lies the problem of trying to anticipate what popular opinion is about a stock. It may not be rational. In the long term, the market does not reward irrationality. It rewards good businesses.
So, stop reading this, and go pick some good businesses to invest in 🙂
Using Artificial Intelligence in portfolio allocation
The term Artificial Intelligence is often misused and abused. What exactly artificial intelligence is, could just as well be a topic of an entire PhD thesis and you’d still not have a conclusion to it. For our purposes, however, let us specify Artificial Intelligence as a machine-driven decision-making system.
Reinforcement learning, one of the many, many techniques used in AI, is essentially training an agent to make decisions in different ‘states’ based on the available action-set. The agent is given a reward for making a good decision or a penalty for making a bad decision. As the agent is trained over and over and over on all the possible states confronted with all the available actions, its decision-making ability improves with the acquired memory of the rewards. Much the same way as us humans learn from our mistakes and become wiser as we get older (at least that’s what I am told :)).
We developed a learning algorithm to train our agent, Meeta, and to get her to answer the popular question of how much to invest in stocks. Her answers were actually quite intriguing. Read more to find out!
Disclaimer : All content published on this newsletter or on any other post on everyfin.in are meant to be for information & education purposes only. It is not intended to be investment advice or a solicitation to buy or sell securities. Please do your own due diligence or consult with your financial advisor before making any investment decision. While the information published on everyfin.in and the newsletters are obtained from reliable sources, neither the author, the publisher nor any of their affiliates guarantee the accuracy or completeness of any such information.