Both machine learning and artificial intelligence have been in the headlines a lot lately; not only in the financial press, but also in mainstream titles. This article from Grant Samuel Funds Management (GSFM) explores machine learning – what it is, how it applies to every day life and importantly, how it applies to investment. GSFM has recently partnered with Man Group to offer Man AHL Alpha, a trend following managed futures fund, to advisers and their clients.
Science fiction has been peppered with heroes and villains that could be described as machines with minds of their own. Wayward computers plotting the overthrow of mankind, or mechanical humanoids intent on destruction or salvation. While the current regime of machine learning has (fortunately) not caught up with all imaginings of science fiction writers, considerable progress has been made in recent years. Some of this has slotted seamlessly into our every day lives, while other advances may take a while to fully integrate.
What is machine learning?
Machine learning is a catch-all name for a range of applied practical algorithms that can identify repeatable patterns and relationships in observed data. Importantly, it can do so without having to be told explicitly what kind of patterns and relationships to look for – the algorithms work that out for themselves.
Machine learning is not a single discipline – it’s a hybrid and borrows from:
- Statistics
- Mathematics
- Engineering/signal processing
- Computer science
All machine learning techniques involve data – the advent of ‘big data’ has given the machines much more to work with.
“Algorithms that can identify and act upon repeatable patterns in observed data; importantly, those algorithms are not told what type of patterns to go and look for – they work that out for themselves.”
Anthony Ledford, Chief Scientist, Man AHL
Although machine learning seems to have hit the headlines more recently, it dates to 1957 when Frank Rosenblatt invented the Perceptron – a machine that could learn to classify images. Today, similar technology is used by a range of programs – a reverse image search in Google can identify the source of most imagery, and Facebook no longer waits for users to tag images, it automatically adds names to photos.
Today’s rapid developments in machine learning are built on three separate and long running revolutions:
Computing power – this has broadly doubled every two years since the 1970s.
Data generation, storage and retrieval – it’s estimated that 90% of the data in existence today were created in the last 2 years; in 1981 one Gb of storage cost $300,000, today the price is below 10 cents[1].
Methodology – practical techniques from statistics, computer science, mathematics and engineering have matured and amalgamated into powerful new algorithms.
Modern machine learning tools offer increased power and flexibility compared to traditional models and purely empirical approaches, and scales well as the dimensionality of the problem increases. This aspect of machine learning is particularly important; as machine learning models are exposed to new data, they can independently adapt. The machines learn from previous experience to produce dependable and repeatable results.
How is machine learning used?
Most people have experienced researching a product or service on the internet at one moment, and finding advertising for said product or service appearing on sites subsequently browsed, or in social media feeds. A field of machine learning known as predictive analytics.
A little more futuristic, but not far away, is the driverless car. There are three things required for a driverless car to work[2]:
- A GPS system
- A system to recognise dynamic road conditions
- A system to turn this information into action.
Both Apple and Google, among others, are working on algorithms to make the driverless car a reality.
Many day-to-day activities in business and leisure are powered by machine learning and continued innovation will see the list of applications grow exponentially.
Examples include:
- Web search engines – start typing and google will start suggesting search options
- Marketing personalisation – personally addressed offers based on browsing or purchase behaviour
- Fraud detection, spam filtering and cyber security measures
- Online applications and assessment for loans and credit cards
- Healthcare, where imaging technology has learned to detect early changes that indicate some cancers and other diseases
- Pattern and image recognition.
Machine learning versus artificial intelligence
Artificial intelligence and machine learning have both grabbed the spotlight in recent times, and the terms are often used interchangeably. They are not, however, the same.
As outlined in the article Artificial Intelligence is a Mega Trend, but who are the beneficiaries?, artificial intelligence is simply taking volumes of unstructured data and plugging it through a ‘big computer’ to give the user a predictive outcome that enhances their experience. In other words, artificial intelligence results in machines being able to carry out tasks in a ‘clever’ way.
Examples include the dynamic news feed on your Facebook profile, predictive shopping results on Amazon, or Google maps highlighting the time your daily trip to work will take, before you asked for it.
Machine learning takes artificial intelligence to the next level – give machines access to the data and let them learn for themselves.
Machine learning and investment
Quantitative or systematic investing has evolved rapidly in recent years, particularly as the opportunities provided by artificial intelligence and machine learning transition from the laboratory into client portfolios.
Machine learning applied within quantitative finance offers a coherent, versatile and practical way of combining numerous and varied weak information sources into investment systems that have greater signalling power than any individual source. Such systems capture insights that both human intelligence and less sophisticated systematic models may miss.
Man AHL has been researching machine learning techniques for five years; it is a core area of research effort both within Man AHL and at the Oxford-Man Institute (OMI), Man AHL’s unique collaboration with the University of Oxford. The University’s Engineering Science Department – of which the OMI is formally part – has a long history of successfully developing real-world applications, such as remote sensor networks and monitoring jet engines in flight. Like others, the OMI is working on driverless car technology. This relationship allows Man AHL to potentially benefit from areas where machine learning methodology from non-finance disciplines can be transferred to quantitative investment.
Examples of machine learning techniques used by Man AHL include:
Deep learning – algorithms using artificial neural networks, designed to mimic the biological networks of a human brain, are trained on large sets of data to ‘recognise’ a range of stimuli. These networks have been used in areas such as image recognition and games such as Go and Chess. They can also be used to learn predictive patterns in financial datasets.
Natural language processing – interpretation of written or spoken language or dialogue. Techniques can be used to assign numerical scores measuring the positive or negative sentiment of company financial reports in a repeatable and unbiased manner. These scores can help feed signals inside trading models.
Quantitative research has typically been predicated on the discovery of linear relationships between input data (such as historical price movements, interest rates or company earnings) and future movements in asset prices. Trend-following is often viewed as a simple linear relationship between past price movements and future ones. If the market went up over some recent time window, it is more likely to keep going up than to go down and vice-versa. If it has been going down in a steep descent, it is more likely to keep going down steeply than if it had been going down only gradually.
What these early machine learning algorithms turned out to be good at is eking out more subtle, non-linear, relationships within data. It showed that within price data, it is not only important that prices went up a certain amount over the last year, the path they took getting there is also important. After three years of trading, and with ongoing research, researchers believe that these kinds of models, when unconstrained, may help identify directional market behaviour including trends, in a way that can be complementary to existing models. As such, Man AHL believes they are clearly applicable to those strategies that seek to benefit from the predictability of market directions.
If this is the age of the data deluge, then machine learning algorithms have the potential to dramatically increase investors’ ability to process and analyse information on markets. There are machine learning algorithms that can help in the decision making around routing between different avenues to market. Man AHL has invested heavily in developing internal algorithms for execution in futures and FX markets.
Applications of machine learning to finance will continue to grow over time. We are entering an age of rapid information growth; data availability is likely to continue to grow by orders of magnitude and vast computing power will be more routinely available. As the field continues to develop, the intelligent application of machine learning has the real potential to help investors capture new and diversifying opportunities in markets.
[1] https://www.ibm.com/developerworks/community/files/form/anonymous/api/library/054c2ab9-ea33-4c70-b0c6-b5bb2482a098/document/7de665ff-2327-41a8b7b05f0bba97356f/media/BIG%20DATA%20%2B%20MAINFRAME.pdf
http://notebooks.com/2011/03/09/hard-drive-prices-over-time-price-per-gb-from-1981-to-2010
[2] http://time.com/3719270/you-asked-how-do-driverless-cars-work/
———–
You must be logged in to post or view comments.