And why it’s nearly impossible to do so
Introduction
It’s been decades since AI has been introduced to the real world and not to mention, it has evolved a lot. AI is now playing a crucial role in almost every field and continues to do in the future too. It has helped scientists or biologists to make breakthroughs in the field of drugs and medicines, environmentalists to maintain and research sustainable environments to mention a few. Most recently, the inventions of GPT-3 and Wudao-2.0, stole the spotlight and grabbed the attention of millions around the world. These are open-source AI technologies that do crazy kinds of stuff like image generation from text, and so on that are beyond one’s imaginations.
Though the current human race managed to discover and enhance such powerful AI technologies that are capable of making major advancements in several fields, it still struggles in forecasting the movement of the stock market. I came across a lot of stories claiming to have predicted the market with Artificial Neural Networks such as RNN, and LSTM and managed to attain an accuracy of 99%. That’s silly and in this article, we will find out why is it so. Considering your curiosity piqued, let’s dive into the article!
Stock Market: The Most Complex Adaptive System
Let’s first discuss what a Complex Adaptive System (CAS) means. CAS is an integrated ecosystem that adapts itself to the given environment. The best example to understand CAS would be our immune system. Whenever a new virus enters the human body, the immune system changes accordingly to be able to respond to the virus. Likewise, the stock market changes concerning the behavior of buyers and sellers.
Now, the stock market being a CAS follows two important characteristics: The first one is, the markets are driven by a network of agents who are nothing but the people. These agents collectively form a system where they interact and follow what the other agents are doing and this cycle loops toward eternity. Secondly, the market’s control is highly scattered. None of the authorities in this world draw the power to control the stock market as a whole. Maybe there exist certain groups to regulate the market but none are authorized to control. So, the overall market relies only on the behavior of millions of investors or traders.
These two characteristics of a Complex Adaptive System might end up being a trouble for AI in predicting the stock prices because the general factor among the characteristics is human behavior and it is obviously inconsistent. First of all, AI can’t take into account human behavior to build a model, and secondly, even it manages to do, human behavior is so inconsistent that it makes the AI model nearly impossible to predict the movement.
AI models can be replicated
Even though when you managed to build a better AI model that actually predicts the market with higher accuracy, it could be imitated easily by others too and that could ultimately alter the future movement of the market you’re trying to predict itself. Let me explain this in detail. Imagine that you have built an AI model to predict the stock prices of Apple and you would be more than excited to make a million bucks out of it in no time. Now, other people can also easily access your model to do the same task, and it becomes a chain reaction. The volume of people using the model and trade accordingly keeps increasing and at one point the total market movement gets affected making the model not be able to predict it and hence becomes obsolete.
Is forecasting the market a waste of time? Probably
According to Warren Buffett, the greatest investor of all time and the father of Value Investing tells that one who claims to have the capability of predicting the future movement of the market is no different than a fortune teller and has nothing to do with investing in stocks. We all know that our investment in the equities will be rewarding in the future and that is predictable, but when precisely this will occur is unpredictable even with AI and other technologies we have at our disposal. What we can do instead is spend our time finding the best stocks to invest our money in and as per Warren Buffett, nothing else gives more assurance to our investment than this.
Is it impossible to predict the market?
To my knowledge, 99% of the people fail in attempting to predict the market but there exist some companies that fall under the category of the rest 1%. Apart from factors such as human behaviors and the market’s random movements, the other important aspect that contributes to reducing the accuracy of the AI model in predicting the market would be bad quality data. So, financial giants like Goldman Sachs and JP Morgan are pouring millions and millions into primarily buying high-quality accurate stock data and secondly into research in the field of AI to optimize their investments substantially.
The best success story to hear from the corporate world would definitely be Renaissance Technology. This company coined the term quantitative finance and always worked toward the conjugation of finance, technology, and mathematics. Their approach is simple. They bought data directly from the exchanges or other high-quality data providers for millions and millions of dollars, fed the data into their mathematical model, deployed the trained model into the real world to predict and time the market. It might sound simple but a lot of complex workings are going behind the scenes and the approach really did pay off. As of now, Renaissance Technologies is the pioneer in the field of applied quantitative finance and one of the biggest companies too in terms of revenue.
The moral of this article would be if you got a hundred million dollars at your disposal, go ahead and predict the market, or else, stop stressing out yourself, grab a bucket of popcorn, and chill with Netflix.