How many of you are interested in stocks? When predicting what company you would like to trade in, you’re probably not in a black box, isolated from the world. Before going for the first company that catches your eye, you have to inform yourself about it, read the news, make predictions as to what to buy and sell. But the market is fast, much faster than humans, so algorithms can be built to make trading decisions for you.
Currently, the default market strategy doesn’t care about the news. It lives in denial about the events around it and just seeks to minimise the spread between the buying and selling price as much as possible. But whoosh here comes a really negative tweet about a company, revealing that they’ve been involved in some scandalous chicken fight, and suddenly loads of people want to sell stocks. If the algorithm doesn’t respond to this it will end up buying stocks from all the people desperate to get rid of them and end up with many stocks of low-value. That’s awful!
Instead, our solution detects how news is going to change the market before it actually does. By leveraging the power of AI models to analyse news we can predict how and for how long a tweet will impact stock prices, updating our trading strategies accordingly.
Microsoft Azure’s Natural Language models were combined with our own feature-extraction algorithms to predict how the stock market would evolve. We then adjusted the bid and ask prices accordingly, weighed by the positivity score of the tweet and incorporated this into our trading bot for the expected duration of the impact. With these strategies we were able to quantify the positivity of each tweet and then gain a notable increase in profit. By incorporating our solution, you can stay ahead of the high frequency money game and maximise the profitability of your stocks.
Log in or sign up for Devpost to join the conversation.