Stockholm (HedgeNordic) – Lars Pehrsson, a software engineer and economist by experience and education, has dedicated the past four years to developing an asset allocation tool that uses algorithms and artificial intelligence to construct and rebalance portfolios of individual stocks. Pehrsson and his team at Robofunds are now preparing to launch their first fund, Robofunds Skandi, which will leverage the tool to build a portfolio of 20 to 30 Nordic stocks. This initial launch is part of a series of planned launches that will employ this tool.
“Initially, we developed an allocation tool for institutional investors and portfolio managers, but later decided to go back to the drawing board and re-design the tool to create a solution we could use ourselves to run investment funds,” explains Lars Pehrsson. “We initially started as a software company providing services to others, but turned around and transformed into a fund management company to use the tool ourselves.” Robofunds continues to offer a white-label solution for investment funds and pension managers that want to make use of the tool powering Robofunds Skandi and the upcoming launches.
“…turned around and transformed into a fund management company to use the tool ourselves.”
The asset or security allocation tool enables end-users to choose an investment universe comprising stocks defined by geographic location, market capitalization, and other parameters. The tool also offers filtering options based on diverse fundamental or market indicators, including price-to-earnings multiples and liquidity, among others. After this filtering process, the tool employs a set of economic models to automatically select stocks with the highest expected return for a given level of risk, as well as facilitates automatic portfolio rebalancing to ensure optimal asset allocation.
Robofunds Skandi
The first fund to use this security selection tool is Robofunds Skandi, which invests solely in Scandinavian stocks. “The first fund invests in Scandinavian stocks and maintains a portfolio that is diversified across all Nordic countries,” starts Pehrsson. From a pool of around 4,000 stocks in the Nordic region, Pehrsson’s tool filters out approximately 300 stocks based on various parameters such as share liquidity, company indebtedness, price-to-earnings multiples, trailing earnings-per-share, and more. This filtering process results in a collection of highly liquid and fundamentally strong companies from different industries.
“Our proprietary algorithm is mostly based on classical portfolio theory, where we do a lot of calculations to minimize volatility and maximize expected returns.”
To determine the expected returns on individual stocks from this collection, Pehrsson employs four traditional methods that consider inter-stock correlations, and volatility, among other factors. “Our proprietary algorithm is mostly based on classical portfolio theory, where we do a lot of calculations to minimize volatility and maximize expected returns,” explains Pehrsson. The algorithm also uses artificial intelligence to conduct optimization calculations for a portfolio of about 25 stocks.
“As an engineer, I first explored artificial intelligence in 2003 and hadn’t touched it until a few years ago,” says Pehrsson. “A few years ago we started studying AI again because of the significant buzz surrounding it when the big driver was the availability of cheap computing power,” he elaborates. For Pehrsson, AI has come into play to address the optimization problem when dealing with a large number of stocks. “If you have two stocks and want to find the optimal allocation, you come up with 101 combinations. If you go to consider five stocks, you get around 53,000 combinations.” The number of potential combinations for a 20-stock portfolio reaches a point where additional computing power becomes necessary. “That is where AI has come in, we use AI as an approximation for traditional optimization.”
“Our tool has an advanced algorithm that considers the trade-off between transaction costs and expected returns when making rebalancing decisions.”
Portfolio monitoring and rebalancing are vital components of Pehrsson’s tool, which is used to run the upcoming launch of Robofunds Skandi. The optimal portfolio is calculated on a daily basis, triggering rebalancing decisions when the existing portfolio becomes suboptimal. “Our tool has an advanced algorithm that considers the trade-off between transaction costs and expected returns when making rebalancing decisions,” emphasizes Pehrsson. “Our tool handles the filtering, optimization calculations, trade execution, portfolio tracking, and rebalancing, and this process goes in a loop.”
While acknowledging the current buzz surrounding AI and Chat GPT, Peehrsson does not view AI as the primary selling point of this fund and tool. “The proof is in the pudding. The true value lies in our ability to find stocks with good returns and low volatility,” says Pehrsson. “Our selling point is our approach to selecting stocks, combined with our dynamic approach to asset weighting and our ability to react within a day’s notice,” he continues. “We have a dynamic approach that rebalances when it is not too expensive to do so.”
Objectives and Plans
The allocation tool also sets an annual volatility target of 15 percent for Robofunds Skandi. As for the return target, Pehrsson says that “no one has ever predicted correctly what the stock prices will be the next day, but our goal is for returns to fall within the range of ten to 15-17 percent per year. No promises, but that’s our goal.”
“We are working on three other alternative investment funds that use the same allocation tool.”
Robofunds Skandi represents the first of a series of planned launches by Pehrsson and his team. “We are working on three other alternative investment funds that use the same allocation tool,” says Pehrsson. The alternative investment fund (AIF) structure provides greater flexibility in terms of cash allocation, portfolio concentration, and the usage of short positions, among other factors.
