Bankers on Wall Street have always tended to look for an edge in old-fashioned ways: a firm handshake, an expensive suit, an invitation for a client to a top sporting event.
But now technology could be about to change the traditional ways of doing things, just as it has in other old-school business lines such as the trading of stocks and bonds. The next frontier is the “primary” side of the banks’ capital markets businesses, which helps companies issue debt and equity.
Bank of America Merrill Lynch, for example, has spent time collecting and cleaning up internal information on initial public offerings it has managed, and fed the 50-gigabyte data set into a machine-learning tool it calls “Predictive Intelligence Analytics Machine”, or Priam.
Since September the bank has used Priam to help predict which investors will be interested in European listings, and it now plans to apply the tool to US and Asian deals too. “Once we’d cleaned all the data, it was unbelievable what we could do,” says Elif Bilgi Zapparoli, the bank’s co-head of global capital markets. “[Priam] is learning and improving with every deal.”
Some senior bankers are dismissive, sneering at efforts to graft artificial intelligence on to what is an essentially relationship-driven business. And even enthusiasts argue that AI will not be a game-changer, given it works best on vast sets of data with billions of observations, rather than the thousands of IPOs and bond deals that banks tend to have in their databases. Humans will always remain at the centre of most of the work, they insist.
Nonetheless, some bankers are optimistic that the often knotty work of arranging debt and equity deals could become much faster and more efficient. Projects under the banner of “digital capital markets” are proliferating across the industry, with banks seeking computer-driven clues on which clients might want to issue, when, and at what maturity and currency. This could result in banks needing fewer people to do the work, at a time when many are still struggling to boost returns to shareholders in an age of low interest rates and tougher post-crisis regulation.
HSBC, for example, recently opened up its internal “MyDeal” app that collects and analyses every titbit of deal information to the bank’s corporate clients. This combines with third-party investment banking products such as WeConvene — which help organise investor meetings and communication — to allow the British bank to streamline every aspect of bond deals, according to Jean-Marc Mercier, global co-head of debt capital markets. He describes the new tech as “like an exoskeleton that allows me to run three times faster”.
Richard Rivero, who heads up a “strats” team within Goldman Sachs’ investment banking unit, points out some of these efforts have been under way for more than a decade. What is different now, he says, is “the volume of data you can collect from our internal processes or external sources . . . and the speed with which you can produce computations”.
He adds: “We are trying to use these techniques to understand how or when companies will come to the capital market to raise funds, to drive targeting for our bankers to help make their coverage more efficient.”
One of the more intriguing efforts to modernise the primary capital markets business is "Project Mars”, a platform for corporate bond issuance that is backed by a consortium of banks including Citigroup, BofA and JPMorgan. It looks set to rival a similar offering from Ipreo, a technology provider, that has attracted several European banks, along with Goldman.
Details are hazy, but people familiar with Mars say it will probably be some kind of online platform that will be integrated with fund managers’ existing order management systems, allowing bond underwriters to relay legal documents, credit rating reports and pricing information to investors, while collecting both indications of interest and firm orders.
That could streamline what is often an unwieldy process. When multiple banks work on deals, say analysts, their salespeople can often end up spamming potential investors with reams of duplicative information and requests through email, Bloomberg chats and calls.
As for BofA, it hopes to use Priam for debt issuance as well as equity, and to use it to predict which covenants — contractual restrictions — have a greater effect on a borrower’s ultimate cost of credit. That could improve efforts to price deals by striking a better balance between supply and demand.
The bank has also produced Insight, a mobile app for corporate clients that can give them detailed breakdowns of market conditions, research, investor demand, and the order book for any deals they are bringing to the market in real time.
“We are transforming ourselves,” says Ms Zapparoli. “It used to be purely about relationships, now it’s about marrying that with data analytics . . . The possibilities are huge, and that’s what’s exciting.”