Machine Learning has the ability to enhance the role of the buy-side trader; bringing trading and portfolio management into a single function.
The trading desk has evolved into a highly specialised function within asset management firms, and they are taking on more portfolio management (PM) responsibilities. Trading desks are now fed with price and liquidity information to support over-the-counter trading, and have highly sophisticated execution capabilities for trading more liquid instruments.
Consequently, traders are “enhanced” in their abilities and, with greater amounts of data flowing, they are also able to feed more information into the risk management and portfolio management functions. This adds greater value to the asset manager. So could an enhanced trader handle the increased complexity of market structure, whilst alleviating the workload of portfolio managers?
There has always been a challenge in separating trading and investment; the PM must integrate information on liquidity and price from the trading process into investment strategies. In addition to close proximity, in order to facilitate communication, more advanced asset managers are able to take the data from the execution process – transaction cost analysis and pre-trade pricing/liquidity data – and use it into the rest of the business.
Clearly this creates real decision-making efficiency. However, the gap could still be closed further. Machine-learning systems are able to take complex data sets – such as the universe of investment-grade corporate bonds for Europe or the US – and alert the trader to relevant dynamics or patterns, such as sudden changes in price, volume or trading direction.
The pressure from regulators and clients for greater transparency is engendering the development of enormous data sets within buy-side firms. The need for consistent and validated data should make the data useful for supporting and training artificially intelligent systems, which can in turn enhance the decision making of traders themselves.
From finding liquidity more quickly to picking the right trading algorithm to use, and assessing viable pricing for investment strategies, AI can be turned to many uses on the trading desk. Looking ahead further, I can see two possible avenues for the trader of the future.
In one, the trader and PM roles are brought back together as a result of the enhanced information flows available to them; greater automation and AI combined allows the handling of a far more complex workload.
In the other, buy-side firms which have excelled in their ability to execute trades are able to provide outsourced execution services to pension funds and asset managers. Some already do this, but with AI supporting a trading desk, the increased efficiency they see will allow them to combine orders from multiple clients more easily, while feeding quantitative reports back to the PMs at those clients.
The role of the buy-side trader is becoming one of designing and applying quantitative models to the market. Buy-side traders engaging with this approach are now building a palpable advantage for themselves and the asset managers they work for. They may in future easily be enhanced and able to handle wearing more than one hat.
The Realization Group and Software AG are running a series of roundtable discussions on putting AI to profitable use in banking and capital markets. Please join us on Wednesday, June 7th to explore making AI and ML work in a live environment.