Qraft Debuts AI ETF to Time Markets
The technology would determine asset allocation based on projected risk level.
Artificial-intelligence-based ETF issuer Qraft Technologies has launched its fourth exchange-traded fund, which it says will help investors time market movements better than they could with traditional indicators.
The QRAFT AI-Pilot U.S. Large Cap Dynamic Beta and Income ETF (AIBD) uses the company’s machine learning algorithm to reallocate between large cap U.S. equity and short-term Treasuries based on predicted risk of market downturn. If the model predicts low downturn risk, assets are allocated to stocks, and at higher risk levels, assets are in short-term Treasuries.
Traditional market-timing measures of risk, such as the VIX, have major downsides, according to Francis Oh, Asia-Pacific CEO of Qraft.
During the market downturn in 2020, for example, the Cboe Volatility Index skyrocketed to 80, but didn’t fall back to a normal reading of around 20 for about a year, he said: “If you’d waited until the VIX has normalized, you’d have stayed out of the market long enough to have missed much of the rebound.”
“When Silicon Valley Bank was falling apart this March, the VIX only moved a day before,” he said, adding the Qraft model signaled two weeks prior.
The actual holdings are fairly simple. Large cap index ETFs like the Vanguard 500 Index Fund (VOO) and the SPDR S&P 500 ETF Trust (SPY) are used for large cap equity exposure, as well as short-term bond ETFs like the iShares Short Treasury Bond ETF (SHV) and the JPMorgan Ultra-Short Income ETF (JPST).
Currently, the fund is 99.9% allocated to the iShares Core S&P 500 ETF (IVV), according to the company’s website.
The launch comes at a turbulent time for some AI ETFs. The machine-learning-based Alpha Intelligent - Large Cap Growth ETF (AILG) and the Alpha Intelligent - Large Cap Value ETF (AILV) stopped trading on May 23.
Oh said Qraft’s fund has the advantage of having a longer track record than most AI-based funds. The specific model for AIBD has been used by the South Korean Hana Life Insurance Company since 2019. He also points to the firm’s large staff of people working on the model: Of Qraft’s 90-person staff, more than 60 are data scientists.
“The funny thing is, three core elements of AI—choice of input data, the architecture of the model and the trading methodology—are all determined by humans,” he said.
Contact Gabe Alpert at [email protected]
