Teucrium Digs New Ground Using AI to Build Commodity ETFs

Teucrium Digs New Ground Using AI to Build Commodity ETFs

Human input remains crucial with new AI-powered commodity ETFs.

Reviewed by: Lisa Barr
Edited by: Lisa Barr

(This article is part of a new series from etf.com highlighting innovation in ETFs.) 

Teucrium wanted to create a commodity ETF that replicated the futures market’s ability to establish long and short positions in certain markets.  

As experienced commodity traders, Teucrium considered creating its own long/short agriculture strategy. But it shelved the idea when it discovered that Singapore-based AiLA Indices had published live results of its long/short commodity strategies using AI since 2017.  

AiLA had impressive results, said Jake Hanley, Teucrium managing director and senior portfolio strategist. AiLA’s long/short agriculture sector strategy index, the AiLA-S033 Market Neutral Absolute Return Index, had a 16.8% average annual return from 2017-2022. The long/short base metals strategy index, the AiLA-S022 Market Neutral Absolute Return Index, had a 15.7% average annual return. 

The Teucrium team wondered if it could do better. “Maybe,” Hanley said. “But why don’t we just license this index instead of building out our own long/short strategy?” In late 2022, the firm launched the Teucrium AiLA Long-Short Agriculture Strategy ETF (OAIA). In April, it launched the Teucrium AiLA Long-Short Base Metals Strategy ETF (OAIB), to complement its half-dozen agriculture-based index ETFs.  

First AI-Powered Commodity ETFs 

The agriculture ETF has exposure to the U.S.-traded futures markets of corn, soybeans, soybean oil, soybean meal, wheat, sugar, Arabica coffee, cotton and cocoa. The base metals ETF has an asset allocation of futures in aluminum, copper, lead, nickel, tin and zinc. Both funds have a 1.49% expense ratio. 

There are other AI-powered ETFs trading equities, but Hanley believes these are the first commodity ETFs to use artificial intelligence in portfolio construction.  

The strategy design of these indexes winnows out irrelevant data, Hanley said, which is why human intelligence matters when using AI for index and ETF construction: “You still need industry experience to know that crude oil is potentially going to be a factor in the price of soybean oil.” 

At a macro level, the indexes incorporate data such as gross domestic product and interest rates. On the micro level, the strategies include commodity spreads, such as a ratio between corn and soybean crop years.  

There are both fundamental and technical components built into the strategies. They include seasonality factors important in commodities, and the momentum factor, although these are not trend-following strategies. Hanley said events such as planting and harvest cycles are helpful for predictive technology to learn about a growing season’s importance to a market.  

Commodity ETFs and AI-Powered Strategies

The ETFs’ objectives are to track the performance of underlying indexes, net of fees and expenses. Teucrium managers execute the trades the machine recommends daily, Hanley said.  

The machine part of the strategy is to eliminate the emotion of trading, and to learn from fundamental and technical factors. That does not mean the machine’s recommendations will always outperform. The agriculture ETF is down 8% this year because the system remained short sugar even as the market rallied. It still is short sugar, he said, and hopes the system will learn from the continued losing position.  

It has been long wheat, and benefited from wheat’s run-up in late July after Russia cancelled the deal to allow Ukrainian grain shipments out of the Black Sea, Hanley noted. 

Other AI-powered ETFs have not had exceptional performance records, as seen by the AI Powered Equity ETF (AIEQ), the Merlyn.AI SectorSurfer Momentum ETF (DUDE) and the Merlyn.AI Bull-Rider Bear-Fighter ETF (WIZ), all large cap growth funds.  

AIEQ is up 4.9% on a five-year basis, while DUDE and WIZ are up 2.1% and 4.1% on a one-year basis, respectively. Those three results compare to the five-year 17.4% annualized gain of in the Invesco QQQ Trust (QQQ) and 12.1% for the SPDR S&P 500 ETF Trust (SPY)

The underperformance in the other ETFs is why strategy design matters in space: “Input is what the machine is actually looking at to make its decisions.” 

Debbie Carlson focuses on investing and the advisor space for U.S. News. She is an internationally published journalist with bylines in publications including Barron's, Chicago Tribune, The Guardian, Financial Advisor, ETF Report, MarketWatch, Reuters, The Wall Street Journal and others.