In less than a decade, AI has gone from an abstraction whose promise pundits debated to a productized technology deployed to address a sprawling array of uses in the finance sector.
That diversity has been instrumental in encouraging the ongoing investment AI requires to realize its transformative potential.
The Ubiquity of AI
Every emergent technology faces a conundrum: To prove transformative, it must influence a broad spectrum of use cases; to influence a broad spectrum of use cases, though, it must prove transformative.
While crypto and blockchain have struggled to clear that bar in the finance sector, AI has sprung over it with sprightly ease.
In fixed income, electronic bond trading platforms like MarketAxess are storming voice trading bastions by developing AI-based tools to price bonds in real time, settle trades, and stymie information leakage. By tracking which traders honor their advertised bids and offers, for example, platforms let their users “see which accounts are honest players and which are not.”
The development is part of a broader movement to rationalize the fixed-income universe — a trend that could one day see machines replace active bond managers.
A May Invesco study of 65 of the largest asset managers over more than a decade revealed that factor exposures explain 66% of bond managers’ excess returns — outperformance AI could conceivably mimic.
Intense margin pressure in asset management could accelerate a turn to AI on multiple fronts. PIMCO is experimenting with machine learning to sort mortgages as well as comb through Fed press releases and meeting minutes for clues to future rate moves.
WisdomTree has meanwhile used AI to identify financial advisors interested in particular products — a refinement that more than tripled cold-call response rates from 20% to about two-thirds.
Like asset managers, banks are deploying AI in diverse roles. UBS is fighting FX flash crashes by using machine learning to “find the bank’s clients the best available liquidity when volatility rises.” The move helped the bank double its algorithmic FX business in 2018.
Banks are leveraging AI to police trading as well as to streamline it. In April, Citi announced it was using AI to sift through more than 9M transactions annually, mapping networks of related parties and spotting customer activity trends to help officers identify noncompliant trades.
Banks are also scouring traders’ emails for suspicious phrases and analyzing computer usage patterns to detect rogue trading behavior.
Success Begets Investment
Collectively, those developments are a sign the diverse benefits of AI are accruing disproportionately to the largest firms.
They are also creating a positive feedback loop that’s encouraging investment in AI more generally. Liquidnet last week announced the acquisition of Prattle, whose AI-based tools “measure sentiment and predict the ‘market impact of publicly available content,’” to bolster its offering of trade and investment analytics.
In March, JPMorgan awarded 47 grants to university faculty and PhD students for AI research. The bank hopes the move will deepen its ties with academia, helping it compete with Silicon Valley in a rush for scarce AI talent.
Such efforts extend even to the national level, where intense competition for AI primacy between the US and China prompted Donald Trump to sign an executive order in February prioritizing AI in its R&D spending. China has already unveiled a similar effort to spur a tenfold increase in AI output in the years ahead.
Taken together, those commitments reveal a virtuous cycle of AI investment and productivity that has all but ensured a radical transformation of financial services in the coming decade.
Tomorrow we’ll dive into the shape that transformation might take in greater detail and highlight some of the challenges AI could face along the way.