Machine Learning in Finance?
Machine learning is an expression we hear more and more in our daily life, and yet, how many of us have a precise idea of what it is?
How many businesses have taken into account the major changes that machine learning and artificial intelligence (AI) are bound to bring to their daily activities?
We’ve all heard about the self-driving Google car and how its engineers are “training” it by driving millions of miles so that it can “learn” all various case scenarios that may occur on the road and eventually become a better driver than a human. Beyond the Go player experiment we all recently have heard of, you may also have read about how AI may save your life by doing a better diagnostic than your own GP. This is simply because it will base its diagnostics not on a few hundred or thousand cases as your doctor would do, but on millions of symptoms that will have been analysed and compared statistically. This AI revolution will have an impact on many aspects of our lives and it will also contribute to transform the financial industry. The noise we hear about the FinTech is still just beginning.
Machine Learning: Become Sherlock Holmes!
So what does Artificial intelligence mean, and especially machine learning, a field of Artificial Intelligence? Imagine you have an almost unlimited capacity to learn and analyse specific sets of data through a statistical approach. You will become like Sherlock Holmes and you’ll be able to solve the most complex crime investigation in no time. An even more advanced machine learning approach allows you to learn from its mistakes without external intervention and now we are in the Terminator movie! Well, all this is still very much fiction but data, statistics and continuous improvement are the key items of machine learning.
Do we need to worry about it?
The “Terminator” image is extreme and we need to keep in mind that AI systems have to learn from data, millions of data well sorted and classified in very specific dataset. This data is not readily available and this is the main protection against the Terminator nightmare. One of the main difficulties of companies involved in Machine learning is to collect and sort the data from which their systems will learn.
As well as being aware of AI, we need to anticipate all the changes that it will trigger in human societies. More and more tasks will be automated as machine will take over all functions that can be “learned”. A huge amount of jobs will, however, be generated around all the machine learning aspects: Data Mining, Data Cleaning and sorting, Mathematics, AI systems training, and also around all the creative aspects (graphic designer, writers…) for which automation will not play a key role at least in the foreseeable future.
Applications in Financial Industry
So, how shall we transpose all this to the financial industry? Machine Learning is already used today in many areas. An obvious and very visible field relates to investment decisions. You have heard about the robot advisors who are taking into account many parameters in order to come up with the “best” investment decisions. Are they, however, using the best datasets to do so? It is difficult to anticipate if a new war is going to break out somewhere in the world and generate a new crisis on the markets. Some could argue that “sentiment analysis” could precisely anticipate this kind of risk, the start of a conflict often follows some specific association of texts and political expressions in various media and a machine could have learned to identify the signals. So, Risk management and compliance controls are other obvious fields for Machine Learning. An AI system can analyse information and patterns from multiple sources in order to detect a suspect behaviour from an investor.
Machine learning can be used for Investments, Risk management, compliance, and also for sales: Financial institutions have so much data available that they could use it to better anticipate their client needs and offer them the right product at the right time.
Finally, machine learning can be used to reduce cost and increase productivity in the financial industry, one specific example is the application of AI to translation so that very technical financial translations can be performed in a much more efficient and quicker way. Lingua Custodia was created in order to apply machine learning to financial translations and help all teams who need to translate financial documents to spend less time and resources on this activity. Whether investment houses use Machine learning to improve their investment performance, to reduce their risk, to improve their sales or to reduce their costs, it should be underlined that an AI system is there to support its “human users”, and certainly not to replace them.
Whether a Fintech or an established institution, the main players of tomorrow in the financial industry will be those who will understand and use machine learning and the AI systems attached to it, in the most optimal manner.