What is the best way to use benchmarks in the hedge fund sleeve of an investment portfolio? There are the classical methods – using fund-of-funds’ performance as a proxy, looking at aggregate benchmarks like HFRI, comparing hedge funds against the performance of public equities – all of which have their limitations. These limitations are in the spotlight now as more investors are giving their hedge fund portfolios a second look.

Hedge fund strategies are back in focus as investors hunt for ways to manage volatility and deal with ongoing distribution issues from other parts of their private markets portfolios, specifically private equity and real estate. Benchmarks are a key part of that discussion as investors look for ways to track performance and use them in strategies like portable alpha which blend a variety of different exposures together to generate returns.

Single-strategy Managers Back in Focus

“Across the board, from endowments, foundations, public pensions there has been more interest in hedge funds recently than perhaps my entire career and they are looking at them in very different ways,” says Elizabeth Burton, managing director at Goldman Sachs Asset Management. She says investors are getting more interested in single-manager funds than they have been in years with a focus on more complex strategies like convertible arbitrage, litigation finance, and insurance-linked securities strategies.

Complex strategies can be harder to assess or measure against a broad benchmark. ILS strategies, for example, might only price their assets biannually. Convertible arbitrage or litigation finance strategies are also going to price their assets at different points and are not easily comparable to a benchmark of listed equities or even a broad bucket of other hedge funds if the benchmark is dominated by components based on, for example, long/short equity funds.

Burton suggests that investors may want to focus less on finding the right comparable index and instead focus on their goals for a given type of exposure and then look at a specific period of time. “If you’re looking at an ILS strategy for example, it’s good to be clear on what function you expect those funds to fulfill in your portfolio – are you looking for longer-term diversification? Then I would look at a representative period of time such as over the last 5 years. How did they perform post catastrophe events or upon triggering events?” she explains. “If you’re focused on shorter-term diversification (such as during a risk off event) you might look at key points like April 2025, to see how the strategy did under pressure.”

Craig Adkins, associate partner at Aon, who leads private credit and hedge fund portfolio management research, agrees. He says it is important for investors to focus on investment goals and create a long-term framework. “I think there is always room for improvement [in performance measurement]. But, do you want to continually change your benchmarking to match the changes of managers?” he asks, adding that creating ever narrower benchmarks and portfolio measurement metrics may not actually lead to better results.

Once you get a baseline understanding of how a strategy holds up for your desired portfolio goal it can be easier to get to a more representative benchmark, Burton adds. “I really encourage people to look under the hood of benchmarks and understand what’s in there to see if it’s actually appropriate for what you’re trying to measure,” she says.

Depending on the goal, there are multiple factors that can influence a benchmark, from the asset size of the included funds, to what strategies it aggregates. Burton says some investors have opted to blend strategy-specific indexes from a provider like HFR together to create a more comprehensive benchmark for their goals.

“The way investors are thinking about hedge funds in their portfolios has changed,” she adds. “There are ongoing denominator effect issues that people are dealing with so they’re looking at liquidity potential. In other cases, investors are looking for diversifiers of their diversifiers, so that’s impacting the types of strategies they’re interested in and also how they are measuring portfolio outcomes. We’re in a period where we’re just seeing a lot of movement in this part of the portfolio.”

Finding the Right Match

A little over a year ago, Michael Shackelford, CIO at the $17.9 billion New Mexico Public Employees Retirement Association, revamped its hedge fund exposure and went through this process. Prior to the change, PERA was using a portable alpha strategy with approximately eight single strategy hedge fund managers. It wasn’t performing all that well.

“The way it was constructed it was using the bond aggregate as the benchmark, so when inflation hit and the Fed started raising rates it slammed bonds. So that index traded off significantly and the portable alpha strategy underperformed as a result of that,” Shackelford explains. “The swaps that were used to generate the beta return were expensive as well, which added to the underperformance and some of the included hedge funds underperformed on top of that. So we opted to unwind that approach.”

Shackelford’s team kept the hedge funds that were performing well, got rid of the managers that were not and replaced them with other single-strategy managers. The portfolio has 10 managers in it now, which is where it will likely stay, Shackelford says. The resulting portfolio, now called absolute return, accounts for approximately 6% of the total portfolio allocation and focuses on strategies that are uncorrelated to the broader market. The team has opted to use the Secured Overnight Financing Rate for overnight secured borrowing+2.5 points as their custom benchmark for performance.

“We’ve chosen that because we went with strategies that are uncorrelated to equities and have very little correlation to the bond market. We feel like if you as a fund manager can’t generate a return equivalent to SOFR plus 2.5 then you aren’t really outperforming,” he explains.

Shackelford adds that benchmarks that use aggregate hedge fund performance like HFRI also include a bunch of underperforming strategies which bring the overall benchmark down in terms of total return. “We just felt like it was sort of the low bar to hit and we wanted to do something else,” he says.

AI Has an Impact

 New research from Tengjia Shu, assistant professor of finance at the University of Illinois – Chicago and Ashish Tiwari, professor of finance at the University of Iowa, looks at how machine learning could support investors in the creation and use of custom benchmarks for measuring hedge fund performance. In a recently updated paper, the professors suggest that by using machine learning investors can parse the factors driving hedge fund performance more effectively and discover new correlations that may not be immediately clear when using more traditional – and lower tech – methods.

To be clear, the researchers are not using ChatGPT and drawing some conclusions. But rather, they are creating new mathematical models and using machine learning to assess performance that includes the risk factors of specific individual hedge fund strategies, historic performance data, and account for issues like the tendency of hedge funds to outperform in a nonlinear fashion as well as hedge fund reactivity to changes in macroeconomic data.

“Hedge funds often exhibit complex relationships to macroeconomic factors,” explains Shu. She highlights how currency momentum and spreads had a big impact on hedge fund performance after the Global Financial Crisis as well as how tail-risk strategies performed during the financial crisis and after. The impact of a single factor on hedge fund performance,  can vary widely and that may not always show up immediately or show up well in traditional benchmarks. AI can account for that in its analysis.

Shu says, going forward, it may be worthwhile for investors to look at using machine learning to get a more representative picture of hedge fund performance. “If you can nail down the factor exposures among the managers in your portfolio, you can tailor your models accordingly, let the model run and analyze the results of that,” she explains. “Our findings show that you can tell which factors were the core drivers of performance over the time period you are examining.”

The models can then be refined over time as new data becomes available, both regarding macroeconomics and at the hedge fund level. The basic framework also allows for evolutions over time, which can support analysis of, for example, how factors interact over the period in question, which could help investors understand latent correlations that impact strategy performance.

“The biggest obstacle that remains is that hedge fund data is not that transparent,” Shu adds. “So a lot of the evolution in our work is happening through process refinements and being able to add more data as we get it. If we start to see more transparent data or at least more current data from hedge fund managers that would improve what we’re putting into the model and therefore measurement. But that’s an ongoing process.”

More on this topic:

	Fixed-Income Benchmarks Become More Customized
	Data, Custom Indexing Reshape Public Equity Portfolios

Tags: Artificial Intelligence, benchmarks, Hedge Funds, HFRI, investment performance, Michael Shackelford, New Mexico Public Employees Retirement Association (PERA), SOFR