In a 2017 lecture, the Deputy Chief Financial Officer of Goldman Sachs, Martin Chavez described how the US cash equities trading division had been reduced from over 600 traders to only two in just 17 years. The cause of what rendered swathes of previously high-paid traders unemployed: automated trading platforms.
Automation and AI are in the process of changing Investment Banking, and their impacts remain unknown. Broadly, there are three main areas where AI is likely to significantly impact: on data collection, trading, and back office.
As any junior analyst or associate will know, a lot of an Investment Banker’s work revolves around data collection. This might entail tracking potential buyers or sellers or drawing up lists of potential investors.
One example of a company automating this process is SourceScrub, which serves as a “private equity deal sourcing platform, an M&A platform and a portfolio tracking platform.” It trawls through 130 million data points across 1.4 million companies, using 50,000 data sources to help its clients, investment banks find targets in the most efficient way.
Analysts are expected to be on top of the financial news 24/7, and AI also has a role to play here. AI platforms can help this information overload, systematising and analysing the news, and using Natural Language Processing to determine its sentiment, allowing analysts to focus on what truly matters to them.
One report from Qualtrics and Amazon found that 97% of researchers felt that AI would render the traditional role of market researcher redundant.
Undoubtedly, many of these innovations will make life a lot easier for investment bankers. Their time can instead be used on high-level, strategic decision-making, as well as networking with peers.
But, while some investment banker analysts might benefit from their resultant higher productivity (and higher compensation), others might find themselves without a job.
AI is revolutionising the trading floor in specific ways. Firstly, through predictive analytics. This is where computer models predict scenarios given particular situations based on past data. One example of this is BNP Paribas’ Smart Chaser, making the trade matching process more efficient using predictive analytics.
Meanwhile, the Man Group has developed routing algorithms to help their traders find matches with brokers and exchanges to fulfil their orders. If those trades fail, this requires labour-intensive exception processing, often requiring phoning up counterparties to discover why the trade fails.
In the future, traders could use “an AI system using a pool of historical data… to identify why trades have failed in the past and apply that logic to new instances,” massively reducing the need for human operators.
On a slightly less back-office note, ING’s Katana software helps traders better price bonds, by using historical and current trading data to help forecast future prices, aiding traders on top of their natural intuitions. As a spun-out company, Katana is now the leading relative value insights tool globally.
In particular, Katana says that, with an accuracy of 91%, it “detects reverting dislocations and reveal relative value insights for up to 200 million bond pairs.” In simpler terms, AI software can sift through the noise of data, find what truly matters, and help traders make better pricing decisions.
AI isn’t just being used in the front office to help with M&A transactions and trading floors, but also with the slightly less sexy – but just as crucial – side of Investment Banking: risk and compliance.
In tackling Financial Crime, AI tools are particularly widespread, such as detecting fraud and anti-money laundering (AML). Moreover, they can often help with data processing, automating the organisation and structuring of messy data sources.
AI tools are also used in Governance, Risk & Compliance, with 42% of respondents indicating its usage in helping with Data validation and 21% on analytical calculations. Here, AI plays a far different role than in the front office: it helps back-office workers do their jobs more efficiently, allowing the clean and organising of data, and the automation of workflow processes, helping identify the extent of Key Performance Indicators (KPIs).
Yet there is a key problem in the case of Back Office automation, specifically with a lack of trained staff. One survey of people working in Financial Crime Risk Management found that 59% of respondents described “Insufficient staff training” as one of the main challenges facing the implementation of Artificial Intelligence.
In short, to get “meaningful results people must understand it”. The lack of trained staff that understand the processes of AI, both in those that are the main users and those they report to, is crucial to the delay in roll-out.
And what does that mean for my career?
Well, it depends on which sector you work. In the back-office, AI processes have not started significantly reducing headcount, instead mainly focusing on improving efficiency and data validation. However, as familiarity increases and staff become up skilled, there might be resultant reductions in headcount.
In the front-office, it depends on which markets you work. As we noted at the start of this article, cash equities have seen a significant reduction in headcounts, while Forex has also been going the same way.
Meanwhile, for investment banking, AI processes may instead allow analysts to focus on more strategic goals and workflows, enabling them to add value more effectively. Whether or not this leads to a reduction in analyst classes at large investment banks has yet to be seen but it may be possible.
However, it is clear that in every sector there are demands for well-trained workers with familiarity and competency of AI processes. For these workers, their productivity is likely to be substantially increased by extensive rollouts, and their bargaining power similarly, at the expense of their less well-trained peers.