In the rearview mirror
It was another quarter of big equities trading results for the bulge bracket banks and some are seeing deal flow pick up. Meanwhile, a survey by Substantive Research showed that most fund managers will move to commission-sharing arrangements. All this should be positive for research.
M&A continued in the sector in Europe. BBVA raised its offer price for Sabadell, Monte dei Paschi di Siena (MPS) completed its acquisition of Mediobanca while Belgian bank KBC is looking at taking over Dutch bank ABN Amro. (For the origin of MPS’s name, see here.)
Research Editors’ SA business is seeing a pickup in terms of client enquiries and interest, and no adverse effects from AI. One big area of demand is the evening slot on the East Coast of the US, which remains a tough one to fill.
SAs on the move
Amanjeet K Chauhan has moved to Desjardins Securities in Toronto as Head of Production from Manulife Wealth & Asset Management – LinkedIn
Kelly Bourque has moved to TD Securities in Toronto as Director, Global Research Business Management from Canaccord Genuity – LinkedIn
Kimberly Williams Manis has joined Canaccord Genuity in New York as Research Production Manager from Wells Fargo where she was a supervisory analyst – LinkedIn
Clint Currie has moved to Stifel in Phoenix, Arizona from Evercore in New York – LinkedIn
Terence Kiernan has joined Barclays from Goldman Sachs in Singapore – LinkedIn
Kevin Chan of China Renaissance in Hong Kong passed his Series 16 in July – congrats to Kevin! LinkedIn
Tay Her Lim has moved to CLSA in Hong Kong as Chief Editor and Quality Control Manager from Exane as supervisory analyst – LinkedIn
Larissa Downes has moved to JPMorgan from China Renaissance in Hong Kong – LinkedIn
Christopher Talay has joined UBS from JPMorgan in Tokyo – LinkedIn
The more it changes, the more it stays the same
A head of editorial remarked to me 20 years ago that analysts need firm boundaries. The sensible ones know that we are the gatekeepers and are the difference between a report going out or not, so it’s better to work with us than against us. But sometimes the pressures of being an analyst get too much. One problem is they are their own franchise and have a limited time to make a name for themselves. If we are the ones blocking their path, it can lead to conflict.
Working from home may have diffused some of this; there is less scope for an analyst to stand over you menacingly demanding a report go out. However, the fundamental requirement to manage them with finesse remains and the best SAs are great people managers as well as wordsmiths. An ex-head of SAs remarked to me the other day that she never said “No” exactly to analysts, more “Great idea. The problem is…” Sound advice.
AI: SAs get a reprieve?
OpenAI published a study recently addressing the issue of hallucinations, which is what happens when a large language model makes up the answer to a question instead of saying it doesn’t know. It said these occur due to the incentives in the models’ training.
LLMs see the prompt/answer dynamic as like a multi-choice Q&A. If ChatGPT does not know the answer and abstains from answering, from its ‘point of view’ it will definitely get 0 marks. If it guesses, it has a 1 in 4 chance of getting full marks. The solution to this problem is for the LLM designers to penalise confident errors (this “guessing”) more than uncertain responses. This makes less likely the scenario of SAs being fully replaced by an “SA report” button in publishing software.
Under current training, with hallucinations incentivised, it would be a brave regulator that allowed a tool that makes stuff up, by definition, to approve research without a human in the middle between it and the analysts. The tool could introduce misleading, false and inflammatory content a rushed analyst didn’t notice, change a headline so that the meaning was lost or introduce a disclosure error. If the models’ incentives are changed, there would still be times when the LLM didn’t know the answer and would have to refer to a human SA to intervene.
In brief
Blue Matrix buys FactSet research authoring tool: Blue Matrix made its third acquisition of the year, buying FactSet’s RMS Partners business. I’ve heard from one client who uses FactSet for publishing that as a result they will be using Blue Matrix from now on.
Fame at last: See this article in The Japan Times about how the independence of research may come under threat for a rare and brief mention of the job done by supervisory analysts.
FINRA published this list of use cases for AI for regulated firms.
Test your knowledge – free practice SA questions from this website.
About us
Research Editors provides research supervisory and editing services to banks, brokers and fund managers across the world.
Our large team of research principals, SAs and editors are based in all major timezones providing round the clock coverage. We have a range of solutions to fit your needs eg running a full department, filling in a difficult to staff position or ad hoc work. Please contact me to discuss.
Disclaimer
Links are provided as a convenience and for informational purposes only; they do not constitute an endorsement or an approval by the author or Research Editors Limited of any of the products, services or opinions of the corporation or organisation or individual. Research Editors Limited bears no responsibility for the accuracy, legality or content of the external site or for that of subsequent links. Contact the external site for answers to questions regarding its content.
Stay updated and subscribe to our newsletter