Robo-finance: The rise of the machines
03 May 2018
David Lewis of FIS discusses how artificial intelligence is driving innovation in the securities finance industry
Image: Shutterstock
When we are in a market that has yet to fully adopt innovations such as central counterparties, such days seem very far off indeed. But, keep in mind, it has taken significantly less than a generation to move almost every aspect of our personal and professional lives online and into an automated or at the very least, a semi-automated environment, so to believe securities finance won’t move in a similar direction is folly. The challenge for the future is about determining where the human element can provide best value in the transaction chain, in conjunction with the machines, rather than resist their introduction.
Those who will have spent a similar number of years in work as I have, may well also remember the training courses run by their employers to introduce email, a fantastic new invention that was set to revolutionise communications. If you are of that vintage, you will probably also remember the fanfare made by Fiat about the fact that the cars they produce were largely built by robots. At the time, it was a desperate attempt to turn around people’s poor perception of their product quality by very publicly removing the fallible human from the production line. Jump forward 20 years or more, and only very special, and very expensive vehicles have any significant level of human interaction as part of their construction. With the advent of driverless vehicles, the car industry has now moved on to replacing the driver in the process, but can robots or artificial intelligence always outperform humans?
When looking at driverless cars compared with humans at the controls, a simple observation of the traffic on our roads would suggest that robots will outperform humans every time. There have been a small number of tragic accidents that have occurred involving self-driving vehicles, but compared to how a human would have reacted to those same situations, the result may well have been the same or worse. What is missing from the equation is the human element, the experience that is hard to teach a microprocessor. The computer, which got its name, in fact, from a human role, is perfect at following rules such as those characterised by physics and mathematics. So, if they are so good at math, are we to expect robots on securities finance desks any time soon?
Looking back to the 1990s and my time at ABN Amro, the then managing director was interviewed on television, as part of a work in the city documentary, making what were quite controversial statements at the time. He was stating that, in his opinion, the day of the “barrow boy” was over and that key personnel would need vastly improved data and a math or physics degree to cope with the way the market, including securities finance, was heading. This was, of course, some time before the arrival of FIS Astec Analytics, and others, bringing transparency of market data to the securities finance arena; but, just as the car industry has moved on, so has data and automation in our businesses.
Over the past 10 years, intraday data has become much more valuable to the market as more of the general collateral, or lower value trades, have become more automated. The advent of new trading venues, plus automation options for low margin, low fee trades, has freed up the human element of the market to concentrate on the higher value trades requiring more individual attention. With increasing constraints being placed on the securities finance market by new regulations demanding greater settlement certainty (CSDR) and heavier balance sheet costs (for example, BASEL), the market has had to become more efficient to continue to produce returns. Regulatory compliance is also a key factor in the automation of services and the increasing reliance on not only data itself but also on the ability to process it and make sense of it. That’s something the European Securities and Markets Authority (ESMA) will have to think a great deal about when it comes to launch Securities Finance Transaction Regulation (SFTR).
FIS clients are using our data like never before; moving away from screens of analytics to greater reliance on superfast application programming interface (API) responses, bringing data directly into the trading process, helping to automate the matching of supply and demand at the right rate, while gathering and delivering the evidence required by regulators to support and validate the best execution. Doing more with less has become a mantra of our business of late, along with many other financial and indeed, non-financial industries. To achieve this and advance, or, as some would argue, just to stay still in increasingly complex and fast-moving markets, more processes need to be automated—as complexities move beyond even the math and science graduates who were envisioned by forward-thinking banks back in the 1990s. Marking a loan transaction, while optimising the returns to beneficial owners, executing a risk-adjusted collateral mix all while complying with settlement and liquidity regulations in the timeframes needed to steal a march on the competition, requires timely, deep, broad and accurate data. The quality of the artificial intelligence is also a fundamental factor. As those experimental, driverless vehicles have shown that computers don’t always get it right.
Last week saw some hedge funds, driven by automated trading, get on the wrong side of a short position as the Automobile Association reported a recovery in its profits and earnings. Those funds that had already banked their gains, following a profits warning in February, and moved from short to long positions ahead of the announcement on 18 April, made some handsome returns. Four of the six funds that publicly reported net short positions (more than 0.5 percent, as dictated by Financial Conduct Authority disclosure thresholds) were computer-driven, algorithmic funds, left at the mercy of a stock that has advanced over 23 percent since the announcement and doubled since the lows of £0.70 a share, seen in March.
It would take a great deal more statistical evidence to prove whether, overall, computer-driven funds were better than the human-managed counterparts, or indeed, whether the next share you lend or borrow will be better value if transacted by a robot. But what is quite likely, is that there will come a time when collateral will be moved by instantaneous transactions recorded in a distributed ledger, servicing lending transactions that settle just as fast and all, probably, before the human in the chain even knew those assets were required at that moment by the underlying algorithmic fund that has created the demand.
When we are in a market that has yet to fully adopt innovations such as central counterparties, such days seem very far off indeed. But, keep in mind, it has taken significantly less than a generation to move almost every aspect of our personal and professional lives online and into an automated or at the very least, a semi-automated environment, so to believe securities finance won’t move in a similar direction is folly. The challenge for the future is about determining where the human element can provide best value in the transaction chain, in conjunction with the machines, rather than resist their introduction.
Those who will have spent a similar number of years in work as I have, may well also remember the training courses run by their employers to introduce email, a fantastic new invention that was set to revolutionise communications. If you are of that vintage, you will probably also remember the fanfare made by Fiat about the fact that the cars they produce were largely built by robots. At the time, it was a desperate attempt to turn around people’s poor perception of their product quality by very publicly removing the fallible human from the production line. Jump forward 20 years or more, and only very special, and very expensive vehicles have any significant level of human interaction as part of their construction. With the advent of driverless vehicles, the car industry has now moved on to replacing the driver in the process, but can robots or artificial intelligence always outperform humans?
When looking at driverless cars compared with humans at the controls, a simple observation of the traffic on our roads would suggest that robots will outperform humans every time. There have been a small number of tragic accidents that have occurred involving self-driving vehicles, but compared to how a human would have reacted to those same situations, the result may well have been the same or worse. What is missing from the equation is the human element, the experience that is hard to teach a microprocessor. The computer, which got its name, in fact, from a human role, is perfect at following rules such as those characterised by physics and mathematics. So, if they are so good at math, are we to expect robots on securities finance desks any time soon?
Looking back to the 1990s and my time at ABN Amro, the then managing director was interviewed on television, as part of a work in the city documentary, making what were quite controversial statements at the time. He was stating that, in his opinion, the day of the “barrow boy” was over and that key personnel would need vastly improved data and a math or physics degree to cope with the way the market, including securities finance, was heading. This was, of course, some time before the arrival of FIS Astec Analytics, and others, bringing transparency of market data to the securities finance arena; but, just as the car industry has moved on, so has data and automation in our businesses.
Over the past 10 years, intraday data has become much more valuable to the market as more of the general collateral, or lower value trades, have become more automated. The advent of new trading venues, plus automation options for low margin, low fee trades, has freed up the human element of the market to concentrate on the higher value trades requiring more individual attention. With increasing constraints being placed on the securities finance market by new regulations demanding greater settlement certainty (CSDR) and heavier balance sheet costs (for example, BASEL), the market has had to become more efficient to continue to produce returns. Regulatory compliance is also a key factor in the automation of services and the increasing reliance on not only data itself but also on the ability to process it and make sense of it. That’s something the European Securities and Markets Authority (ESMA) will have to think a great deal about when it comes to launch Securities Finance Transaction Regulation (SFTR).
FIS clients are using our data like never before; moving away from screens of analytics to greater reliance on superfast application programming interface (API) responses, bringing data directly into the trading process, helping to automate the matching of supply and demand at the right rate, while gathering and delivering the evidence required by regulators to support and validate the best execution. Doing more with less has become a mantra of our business of late, along with many other financial and indeed, non-financial industries. To achieve this and advance, or, as some would argue, just to stay still in increasingly complex and fast-moving markets, more processes need to be automated—as complexities move beyond even the math and science graduates who were envisioned by forward-thinking banks back in the 1990s. Marking a loan transaction, while optimising the returns to beneficial owners, executing a risk-adjusted collateral mix all while complying with settlement and liquidity regulations in the timeframes needed to steal a march on the competition, requires timely, deep, broad and accurate data. The quality of the artificial intelligence is also a fundamental factor. As those experimental, driverless vehicles have shown that computers don’t always get it right.
Last week saw some hedge funds, driven by automated trading, get on the wrong side of a short position as the Automobile Association reported a recovery in its profits and earnings. Those funds that had already banked their gains, following a profits warning in February, and moved from short to long positions ahead of the announcement on 18 April, made some handsome returns. Four of the six funds that publicly reported net short positions (more than 0.5 percent, as dictated by Financial Conduct Authority disclosure thresholds) were computer-driven, algorithmic funds, left at the mercy of a stock that has advanced over 23 percent since the announcement and doubled since the lows of £0.70 a share, seen in March.
It would take a great deal more statistical evidence to prove whether, overall, computer-driven funds were better than the human-managed counterparts, or indeed, whether the next share you lend or borrow will be better value if transacted by a robot. But what is quite likely, is that there will come a time when collateral will be moved by instantaneous transactions recorded in a distributed ledger, servicing lending transactions that settle just as fast and all, probably, before the human in the chain even knew those assets were required at that moment by the underlying algorithmic fund that has created the demand.
When we are in a market that has yet to fully adopt innovations such as central counterparties, such days seem very far off indeed. But, keep in mind, it has taken significantly less than a generation to move almost every aspect of our personal and professional lives online and into an automated or at the very least, a semi-automated environment, so to believe securities finance won’t move in a similar direction is folly. The challenge for the future is about determining where the human element can provide best value in the transaction chain, in conjunction with the machines, rather than resist their introduction.
NO FEE, NO RISK
100% ON RETURNS If you invest in only one securities finance news source this year, make sure it is your free subscription to Securities Finance Times
100% ON RETURNS If you invest in only one securities finance news source this year, make sure it is your free subscription to Securities Finance Times
