L-R: Amanda White (Conexus Financial), Judy Wade, Dianne Sandoval, Mohan Balachandran, Matthew Shellenberger. Image: Jack Smith.

The $87 billion HESTA is using artificial intelligence to enhance its multi-asset quant strategy overseen by head of portfolio design, Dianne Sandoval.  

But operating in the highly regulated Australian market means being able to explain how the fund arrived at investment conclusions via these models is almost as crucial as generating alpha, Sandoval said. 

During a panel at the Fiduciary Investors Symposium at Stanford University, hosted by Investment Magazine sister publication Top1000funds, Sandoval explained that two AI applications in HESTA’s quant program are neural networks, used for equity forecasts; and advanced statistical tools to explain the behaviours of those models, including conducting Shapley decomposition analysis. 

“We have to be really careful about ensuring that we have KPIs and audits,” she told the symposium. 

“It’s really important that we can be really clear as to what the model is telling us, why the model is behaving the way it’s behaving, and that – if we get audited or the regulator comes in – we can explain the behaviour of that model and it’s not a black box.” 

HESTA’s signals have an 18-month horizon, she said, and aim for a CPI plus 3 per cent goal over the long term. 

“This program has added 83 basis points of alpha, and has been very successful,” Sandoval said. 

“The thing I think that has made it so successful is that it tends to be counter-cyclical, so we’re essentially diversifying some of the active return streams that our active managers are doing. 

“Through equity drawdowns, this this program has actually added alpha.” 

Success metrics

Sandoval said the program, like everything else the fund does, comes down to the nexus of three things: performance, cost and suitability for members. The symposium also heard how other US and Canadian pension and endowment funds on the panel have set barometers for success when it comes to AI usage in their respective organisations. 

For Mohan Balachandran, senior managing director of multi-asset strategies group at Teacher Retirement System of Texas (TRS), one straightforward metric is performance. He oversees TRS’ quant equity and quant macro programs, with the former being a US$21 billion ($30 billion) long-short portfolio and the latter being effectively an internal hedge fund focused on cross assets in commodities, FX and bonds. 

The biggest use of AI in Balachandran’s team is decision trees for picking up signals, he said. 

“In the equity portfolios, it used to take a long time to update your models, because people would do univariate regressions and things like that,” he said. 

“But once we started using this, it’s just become a very quick and fast turnaround, so much so that now instead of global models, we’re more focused on sector models and country models.” 

Another plus is that the models are usually open source, Balachandran said, which means the fund’s cost is limited to the data side. 

TRS tracks its internal quant equity program’s information ratio against those of the hedge funds it invests in, and as of February 2024 its internal program has outperformed seven out of ten external funds. 

“If we update this for the end of June, we’re leading this pack, actually,” Balachandran added. 

“I have 12 people in my team who built this [portfolio] out, whereas if you compare it to any of those hedge funds there, you can add another zero to the number of people working on those products. 

“There’s lots of knowledge workers [in financial services], and there’s lots of data, but what I find is that we can do a lot of that in-house with a lot fewer people [with AI].” 

Internal focus

CPP Investment’s Judy Wade has more of an internal focus when she considers AI’s success metric. Wade heads up the fund’s San Francisco office and is head of strategy execution and relationship management, and as a part of her broad remit led the development of a generative AI knowledge platform in the fund. 

On the so-called “first mile of investing”, Wade said successful technology integration should be able to drive efficiency and access to do research or respond to investments. 

“On the last mile investing, we’re just starting to do some experiments that can really help investors. For example, trying to predict and looking at sentiment analysis more quickly,” she said. 

“Taking the fact that we have ingested all of our fund’s PE managers’ materials, and our venture growth materials – if you’re now in the public markets, can you find disruption risk that you wouldn’t otherwise.” 

On the spectrum of being a “taker, maker or shaper” of technology, Wade said CPP is somewhere in the middle. 

“[We] imagine it as both a navigator and a workbench for an investor,” she said. 

“We do fundamentally believe that it is our data combined with large language models that provides us with the proprietary insights – that is our data and our partners’ data. 

“As part of that, we really fundamentally believe in attribution, we really believe that if you get something back from our platform, you should be able to see the exact sources…and our partners care about that too.” 

WK Kellogg Foundation senior manager of asset allocation and risk management Matthew Shellenberger echoed AI’s role in administrative excellence but said its impact is more acutely felt on a smaller team, such as the foundation’s. 

The foundation has more than 100 manager relationships and Shellenberger said it is utilising machine learning tools to scrape data, summarise documents and support its accounting system. 

“We’re 11 [team members] overall, but being able to digest materials from over, let’s say, 100 relationships, and actually be able to cultivate our own portfolio and our own tilts, I’d say those three components are quite valuable for us.” 

“The one other part of this [AI]…is it allows us to do what we really love. Most of us got into the field because we love investing, and we’re infatuated with that idea. 

“By having these tools that basically allow you either during a meeting to focus more on the active conversation…or distilling notes over the course of a year into a nice executive summary for your reporting structure – whether it be your IC or board – it allows you to be the investor that you grew up wanting to be.” 

Join the discussion