For years, sophisticated asset managers gained an edge thanks to unique intel that didn’t come from traditional market sources like stock exchanges. Now, the advantage is inside their own firms.
Hedge funds and others were better able to predict quarterly earnings and commodity prices thanks to so-called alternative data from aggregated credit-card receipts, retail foot traffic tracked by cellphones, and satellite images of crop fields.
Then these sources became so widely used across the asset management space that they became table stakes rather than differentiators. So asset managers pumped resources into finding new, obscure data sources that might help them beat their peers across the street.
Now, the rise of large language models has eroded that edge even further. With top funds hoovering up as much publicly available data as possible with AI, investors say the next market-beating edge will come from searching their own research, communications, and historical decisions for signals competitors can’t access.
AI is “great at structuring unstructured data,” said Jacob Bowers, a vice president of quantitative research at BlackRock, on a panel at the Future Alpha conference in New York on Tuesday, and “some of the best unstructured data you have is internal.”
The publicly accessible data that was once cutting-edge is now “commoditized” by AI, he said. BlackRock, the world’s largest asset manager with $14 trillion in assets, has already turned its agents internal to find potential investment signals within past communication between investment professionals and old reports on opportunities, he said.
The gold mine of data within long-running asset managers is well known. A 2019 report from consultancy Opimas said it expected funds to eventually sell some of their data to generate additional revenue, and Robert Frey, a former Renaissance Technologies managing director who runs a fund of funds, told Business Insider then that his former employer’s biggest advantage was its “massive data library” gathered over decades of trading.
And AI has made it much easier for funds to tap into this potential source of alpha.
Andrew Gelfand, a quant at Balyasny focused on alpha capture, said at the Future Alpha conference that the firm had previously tried to monetize unstructured data within the firm’s systems, but recent AI advances have made the task much more fruitful.
The $33 billion firm requires analysts to type their research and notes into a portal that his team can access, Gelfand said, giving AI reams of text to sift through for potential investment signals.
What this type of internal mining requires is top-notch data for AIs to learn from; in other words, the thoughts and processes of seasoned investors at the top of their game.
While funds may have more internal data than any human could process in a lifetime, AI agents constantly need more information to keep up with a changing world and evolving markets.
“You need the feedstock to be high quality,” said Mike Daylamani, who runs a team blending fundamental and systematic investing at Engineers Gate, at the conference, referring to the data feeds quants use to build their models.
“At the end of the day, this is a creative endeavor,” he added about investing generally.
Read the full article here
