Use of AI in Stock Markets
The performance of Capital markets is a very strong indicator of a country’s economy. One of the Capital markets is the Stock market, which is powerful, yet volatile but as said, very vital. The securities traded at a stock exchange can decide the future of an individual or an investor and then holistically the trajectory of a country’s economy. With the advancement of today’s technology the use of Artificial Intelligence in stock trading had been inevitable and currently being used globally. According to Forbs and Wired, AI are widely used in stock trading and currently at least 1,300 hedge funds are using some sort of AI models to decide their trade. And many more.
Therefore I will elaborate on two implementation of AI in stock markets domain in this article. They are 1) Predicting the Stock Market volatility and 2) use of Intelligent Agents for trading stocks.
Let’s dive in,
1. Stock Market Volatility Prediction
Volatility in a stock market is the rate of rise and fall of stocks given a particular time period. Knowing the market volatility helps to make policy decisions on the market, portfolio rebalancing, hedging and many more. Using ARCH and GARCH Models, estimation of volatility had been done. After that in the industry Volatility forecasting had being conducted using ANNs (Artificial Neural Networks).
In a study done by FCS New York had seen the use of Clustering for volatility prediction by using DI in Indian stocks. This study had used 3 algorithms,
- Kernal K-means algorithms
- Self Organizing Map
- Gaussians Mixture Model
Then by varying the number of clusters and variables the Silhouette index was checked. The study has found that lesser the number of clusters the better the use case is. The two years data of Indian stocks had seen 7 predictors given 5 to 6 clusters providing the optimum results in the Study.
2. Trading in Stock Exchange using Intelligent Agents
In reinforcement learning we are having intelligent agents who learns by them selves. For this in reinforcement learning a reward mechanism is in place where for an act a reward is given. Based on the rewards this agent learns in an environment.
Below is an example of Intelligent Agent acting as a Trader.
In the example, the stock prices are sent to the Agent as observations from the Market. Which is also the environment. Then the agent, who is a neural network will go through the data and past learning to come with the actions. Actions are whether to Buy, Hold or Sell the stocks. After deciding, this action will be implied to the environment which is the Trade Market. when the trading is done, data which tell whether the trading is a profit or a loss, which will again get injected to the Agent as the reward. This is a sequential activity where the target at the end of the day is to come out with the maximum profit.
Hence stock trading does not only affected by the stock prices, there are also many other external stimuli. These can be what’s happening in the country or in the world, tweets by influential figure and even what’s trending at the time. An instance or some times a single NEWS can even affect the trends or how the stock market will move. Below is a diagram showing, how all of these external stimuli can be ingested to the Agents decision.
According to the above diagram Data from stocks, Textual data and Google trends are integrated together. Then by partitioning them into Training and Testing sets, an AI model can be created. When getting textual data, some data pre processing and text mining should be done. A good opportunity is also present to carry out a sentimental analysis.
The development of such kind of an Intelligent agent would be a time consuming and difficult task, but it is possible with the advance technology we are currently having.
Trending Globally…
According to Coalition, a U.K. research firm, 45% of revenue in cash equity trading had being resulted in electronic trading. Data are enormous. Nowadays even Hedge funds use AI related analysis for their investment ideas and to build their portfolio.
There are many global tech companies currently using AI in the Capital Markets domain. One of them is Trading Technologies which is a Chicago based software company developing AI platform that can identify complex trading patterns in a real time setup on a massive scale across multiple markets. Then there is Auquan, a company developing a platform where data scientists can create Algorithmic trading strategies, and had won the award for the Hottest Fintech in Europe at the 2019 Europa awards.
Trade Ideas is another San Diego based company which outperformed the market benchmarks in first quarter of 2018, to a returning of 16% to the S&P’s -1.0% that quarter, which shows the power of AI in action. And the list goes on. Also this trend in global startups and IT firms to adopt latest advancements in AI to dive into capital markets had been a resounding success, where the intentions are clear, which is more accurate and quick predictions.
“We’re basically a data science platform for investment management. And we’re trying to make the process involved more efficient, more accurate, and better by incorporating alpha-generating insights extracted from large amounts of structured or unstructured data,”
— Chandini Jain (Founder and CEO of Auquan)
So you can see, that the use of AI is not a concept but a widely used reality in stock markets and the Capital market domain. Therefore we can say that technology wise global advancement in tech in this area is very commendable. So now it’s our turn to use the tech which are being developed in this competitive market for our investment goals. Also this tech is a good chance for the aspiring startups in Sri Lanka to step into the Capital Markets domain to test the waves. Yes it is competitive globally, but most importantly, it is possible. Where the real magic is!
