From T+1 to Artificial Intelligence

Some of you may have noticed that I didn’t have a chance to publish for the past couple of days.  I was far from idle, however.  I was fortunate to be asked to speak at a major conference.  The attendees on the first day were primarily Chief Investment Officers (CIOs) at major institutions, while the morning of the second brought a large group of Chief Technology Officers (CTOs) to the mix.  Most of the first day’s topics involved outlooks for economics and earnings, along with techniques and best practices for senior portfolio managers; the second day was focused heavily on how investors might incorporate artificial intelligence into their investing processes.  I’ll summarize some of them later in this piece.

Yet amidst all these high-level topics, my specific role was to discuss the upcoming transition to T+1 settlement.  As I put it, we heard from a series of architects, city planners and landscape designers who described the big picture for large investors.  Our panel was the meeting with the plumber and electrician who have to tell you whether those designs will work as planned and whether they will meet your budget.[i]

For those unfamiliar, there will be a change coming to the settlement cycles for North American equities.  Normal settlement for US and Canadian stocks currently occurs two days after a trade is executed, or T+2.  On May 28^th, that time period will shrink so that stocks settle on the day after a trade, or T+1[ii].  FINRA recently put together an article explaining how the change should affect average investors.  The short answer is that most investors won’t notice the change, particularly if you are an individual or small institution that trades and clears here at Interactive Brokers.  (And if you’re not – what are you waiting for?). 

Large institutions may have other issues though.  The change should affect the speed at which large investors inform their prime brokers about how they allocate executions to subaccounts.  The change could also affect the availability of stock borrowing, as stock-loan departments adjust to less notice for availability and recall.  Also, since most global markets and currencies will continue to settle T+2, that could create some mismatches.   Those should be all temporary speed bumps for the industry, but if you are a fund manager who engages routinely in stock lending/borrowing, have lots of institutional subaccounts, or global assets and customers, you should be carefully examining your policies and procedures.

Now onto the sexier part.  Conference attendees got to hear from a range of cutting-edge experts about how they are utilizing artificial intelligence in the investment process.  The best current use cases are for AI to automate several parts of an institutional portfolio manager’s or analyst’s job that are currently time consuming and task specific.  Those could be sifting through news reports, analyst calls, and corporate disclosures to compile news items that an investor might find relevant.  AI can indeed speed up those necessary but time-consuming activities, enabling more time to be spent interpreting the data rather than simply entering it.  AI can also help speed up coding, something that is crucial, but not specific to investing.  Yet we heard little about how AI is revolutionizing the investment decision process. 

It would be understandable if few managers would be willing to disclose that in a public forum, but the promise of AI so far appears to be in facilitating the tasks necessary for making intelligent investments, not making the investments themselves.

Amidst the enthusiasm, one panelist’s offhand comment stuck most with me.  He said something to the effect of, “remember how many panels we had a couple of years ago about blockchain’s use in investing?”  I didn’t take it to mean that artificial intelligence had little to no role in the investment process – even if blockchain has little role now – but it did put the topic into perspective.  AI is THE market obsession right now.  It’s driving the performance of several key stocks and therefore the equity indices that they inhabit, even if its current uses seem more evolutionary than revolutionary.

Can AI meet the hype?  Over time, possibly, if not probably.  Remember, the internet eventually lived up to, if not exceeded, its 1999 hype.  But many of the perceived leaders failed or were surpassed.  Cisco Systems (CSCO) was the Nvidia (NVDA) of its day.  It took until 2022 for its stock to finally surpass its 2000 high.  Consider that Meta Platforms (META), a company that didn’t exist during the internet bubble, added about $200 million to its market capitalization last Friday.  That’s about the current market capitalization of CSCO.  Put differently, META added a CSCO on one post-earnings rally.  AI may indeed deliver on its promise, but what if it takes longer than expected?

—

[i] Personally, it was gratifying when more than one attendee told me that I made a dry topic interesting and brought a sense of urgency about a change that could affect many of the CIOs in the room.

[ii] By the way, options already settle T+1.