(MENAFN- ValueWalk) London-based Man AHL has seen the machine learning future and it is more a 3-D experience than black and white. One of the early trend following adaptors – who is credited with spawning the likes of Sir David Harding of Winton Capital, the H in AHL – Man AHL is an heir to derivatives industry royalty. But one of the earlyis also making headlines as it works on the cutting edge of machine learning. The firm was the first known CTA strategy practitioners to have turned machine learning into a , CNBC's Leslie Pickar and Juliaane Funk first . In a September note to investors, the firm's portfolio strategy head, Graham Robertson, teams with portfolio manager Mark Refermat to explain the innards of what might be categorized as the first successful application of machine learning in trading. What they have discovered to date both confirms the validity of the primary strategy that underlies Man AHL but also points to new correlation patterns and .
Get The Full Ray Dalio Series in PDF Get the entire 10-part series on Ray Dalio in PDF. Save it to your desktop, read it on your tablet, or email to your colleagues
Could James Man have forecast the trend of machine learning?
When James Man was operating a commodities brokerage near London's roughened shipping docks in 1793, supplying rum to the Royal Navy, little did he know the firm would spawn the world's publicly traded hedge fund. It wasn't a shock when the storied derivatives firm would, in 2007, split into MF Global for the brokerage business and Man AHL for the investment business. A nasty argument over transactional fees and an inferiority complex was said to be at the heart of the fight, with a legal battle documenting the less than cordial relationship. What was a shock is that the firm survived the primary thought leaders walking out of the business in 1996 – the A, H & L in the name, Adam, Harding and Lueck, all left the firm to start their own business ventures. This firm that has lived through so much of derivatives industry history has survived to lead a revolution where computers are replacing human thought processes and creating entirely new trading systems.
It is under this industry milestone that new names are being added to the legend. But this time Robertson and Refermat -- the future R & R Capital? -- are giving billing to machine learning for creating the latest innovations in financial services, not just human minds.
'Machine learning helps us see the world through a different, and not always linear, lens,' the pair wrote in a robotically titled piece 'Machine learning in investment management.'
To recognize the difference in outlook between the historically linear outlook of the world is to understand how CTA strategies have historically operated. The math formulas for trend following, for instance, can all have different nuance, but there is one variable that doesn't change: how they respond to a beta market environment of price persistence. While formulas can capture this technical market environment differently, the methods to accomplish this have always been centered on if / then mathematical logic. Said another way, most formulas measured the strength of market buying and selling to determine when a certain threshold was crossed, a point at which a trend following trade signal to buy or sell was generated. If / then Boolean logic drove math formulas in a linear direction.
Robertson and Refermat have their take on the if / then nature, and then make an interesting discovery:
When building machine learning algorithms to identify the best trading patterns, the computer kept returning the same result:
Because this relationship has been so strong in the past, it is not surprising that, when left to their own devices, many machine learning algorithms 'discover' trend-following as the first way to forecast future prices when given, as inputs, historical market prices.
The concept of why market trends exist from a fundamental standpoint has been credited, in part, to behavioral economist and University of Chicago Professor . In part, his Nobel Prize-winning work recognized that market behavior is illogical and different human market participants recognize different market inputs at different points in time. While Thayler , the same is true with algorithms recognizing a trend or market movement at different points in time.
Robertson and Refermat, while on the cutting edge of artificial intelligence and machine learning, have figured out one commonality of a strong algorithm, a bit of market wisdom voiced by many of the best fundamental managers in the world as well.
'There is no value in trading a more complex model if a simpler one is just as good,' they noted. In reality, most quantitative managers can explain a strategy to a fellow insider in less than 10 minutes when following a beta market environment protocal. Its when client pitches and public explanations come into play that the obfuscation tends to fly. Robertson and Refermat, in their clearly written note to investor, avoid the obfuscation and point to a machine learning future that is about best execution along with trade ideas.
After three years of trading, and with ongoing , we believe that these kinds of models, when unconstrained, may help identify directional market behavior including trends, in a way that can be complementary to existing models. As such, we believe they are clearly applicable to all our strategies that seek to benefit from the predictability of market directions.
It is in the 'predictability' comment that Robertson and Refermat are differing from in the trend following industry who say a market top or bottom cannot be predicted; trend followers just follow a trend after it has been established.
Mountains of computer data is analyzed to make machine learning work
MENAFN1710201701980000ID1095962076
Legal Disclaimer:
MENAFN provides the information “as is” without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the provider above.