One of the reasons why many never really get into trading stocks and shares is because it’s complicated. The process and concept itself are relatively easy, each share has a value which goes up or down depending on market conditions, but in a world where there are tens of thousands of companies to invest in, how is it possible to ever know what the market is saying?
People like Warren Buffett seemed to have an innate understanding of where the market is at any one point, but in reality, there is no way for a human being to know if a company is going to announce something big or have a negative incident shared across the internet. For instance, in early April United Airlines saw their share price take a hit thanks to a video appearing on social media showing a passenger being forcibly removed from a flight. For those tracking the markets, a video appearing on Facebook is not within any kind of regular trading algorithm’s remit, but through an increasing use of big data, companies can include data from a huge range of places.
June 2015 marked a big shift in this direction, with Thomson Reuters reporting that it was at this point that they had more customers who were machines than humans. The financial information service was, therefore, providing more information to be fed directly into algorithms than for humans to make their own informed decisions. We are likely to see this trend continue. With financial services and banks regularly amongst the least trusted institutions in the world thanks to headlines surrounding scandals like the rigging of the libor rate and films portraying true stories of recklessness like The Wolf of Wall Street and The Big Short, it seems prudent to make these decisions based on more than just gut feeling alone.
The spread of big data and the speed in which changes can occur to share price based on a huge number of factors is perfectly suited to big data, AI, and machine learning. We have seen how Tweets from specific people can impact share people and net or lose people millions, such as Donald Trump’s tweet criticized Toyota saw the car maker lose $1.2 billion in value. It has led to some journalists from NPR even creating algorithms based purely on the relatively erratic public statements of the president, to see if they can take advantage of this phenomena. However, if somebody else with fewer followers were to tweet that they were unhappy with their Toyota, it is unlikely to have any impact on the company’s share price.
However, as we saw with the United video, which was posted by an innocuous person with a tiny fraction of the influence of Trump, viral posts can come from anywhere and have equally damaging effects. Data technologies allow people to know about a post potentially going viral about a company before a journalist has a chance to spread it to the world and create the kind of damage seen by United in early April. The real difficulty of investing is that it is largely illogical, with markets reacting to different changes in unpredictable ways. Scanning data from social media platforms has been shown to be more accurate than traditional modeling, with a team from the University of California creating a model in 2012 that scanned Twitter for mentions of companies which included sentiment analysis of what was said, which was 11% more accurate than traditional models.
This suggests that the power of the message on social media may actually have more of an impact on share prices than anything to do with hard evidence. There have even been studies that show the excitement or deflation of the sentiment of chatrooms talking about a specific share can have an impact on the price of that share. Sentiment analysis and machine learning can learn to understand when this is taking place, then either automatically take action to buy or sell that share, not based on concrete evidence, but on the sentiment surrounding it.
Traditional brokerage firms are increasingly under pressure to not only provide large returns but following the 2008 financial crisis, to also minimize risk. This, alongside the trust placed in data by the general public, could be one of the reasons why the increased use of big data in modeling. It fits with what Andre Cappon wrote in Forbes, that ’institutional brokerage as a business is under pressure – the industry is likely to continue consolidating. The winners will be those firms that specialize and innovate.’ The use of big data technologies does both, with the companies collecting and effectively analyzing the most data the least likely to be caught unaware and lose money and the most likely to be seen as technologically innovative.
However, it is not only the big trading houses who are getting involved in this through their traders, the kind of information available only to trading houses is now available to almost anybody. There has been an awakening of the general population, with close to 50% of all amateur investors now taking a ‘do it yourself’ approach. This has created a new market for ‘robo-advisors’ provided at relatively low cost, but still allowing some of the established players to keep a connection to this growing segment who have largely shunned traditional investment companies. These advisors give information to DIY traders to help inform their decisions, with companies likely to start competing more intensely to provide the best information to attract more of this large DIY investor segment. This will likely see more and more big data technology being included in these robo-advisors, bringing them closer to those used by professionals in the future.
Big data is currently having a huge impact on investment, and with the constant search for an edge from investment companies, is likely to be one of the most important technologies for future growth. It used to be that the quants advised the traders, now it seems that the quants could well be taking over entirely.