Is there an AI Index Fund? What is the performance of Ai Managed ETF’s?
Today let’s take a look at the Ai ETF offering. And their performance.
Within the past decades, AI has been extensively utilized in the financial market. To be more specific, AI becomes employed in exchange-traded products. Including exchange-traded funds (ETFs), exchange-traded notes (ETNs), and other similar product types transacted in the market.
Generally, ETFs are pooled investment options consisting of baskets of securities. Including stocks, bonds, and other assets, organized according to the stated fund goals. AI ETFs benefit from adopting artificial intelligence because AI will lead to a wide range of chosen stocks with improved customization and affordability.
According to ETF Database, there are three primary criteria for categorizing AI ETFs, as shown below:
- FIrstly, they are funds that specifically invest in companies developing new products or services and technological improvements in scientific research related to artificial intelligence.
- In addition, funds with at least 25% of portfolio exposure to companies that spend large amounts on artificial intelligence research and development (R&D) expenses.
- Lastly, funds that use artificial intelligence methodologies to select individual securities for inclusion into the fund.
An ETF needs to fulfill either criterion to be characterized as an AI ETF. This article focuses on the ETFs that satisfy the 3rd measurement because it is crucial to determine how artificial intelligence has impacted ETFs’ stock selection processes and performances.
After preliminary filtered research, 6 ETFs use AI as part of their stock-picking strategies, and their major components are U.S. stocks. They are listed below in ascending order of inception dates:
- WisdomTree U.S. AI Enhanced Value Fund (AIVL), issued by WisdomTree on Jun 16, 2006
- AI Powered Equity ETF (AIEQ), issued by ETFMG on Oct 17, 2017
- SPDR S&P Kensho New Economies Composite ETF (KOMP), issued by State Street on Oct 22, 2018
- QRAFT AI-Enhanced U.S. Large Cap Momentum ETF (AMOM), incepted on May 21, 2019
- QRAFT AI Enhanced U.S. Large Cap ETF (QRFT), incepted on May 21, 2019
- QRAFT AI-Enhanced US Next Value ETF (NVQ), incepted on Dec 02, 2020
The evaluation of these 6 AI ETFs is based on efficiency, tradability, and fit. The efficiency metric is about whether the AI ETFs fulfill their core promise to investors without incurring excessive fees or risks. It has three main category variables: expense ratios, performances, and risks. The tradability measures if investors are capable of getting fair deals buying or selling in the open market. It includes two factors: trading volume and bid/ask spreads. In addition to the efficiency and tradability characteristics, the fit metric provides insights into how AI ETFs perform in markets relative to their benchmarks. This incorporates two measurements, downside/upside capture ratios, and Sharpe and information ratios.
Efficiency Measurement: Expense Ratios
The first component of the efficiency measurement is the expense ratio. By researching the AI ETFs’ prospectus, the following graph demonstrates their expense ratios and their expense ratios versus those of the Morningstar category average.
The expense ratio indicates how much of an asset of a fund is used for administrative and other operating expenses. Thus, the cost of investing in KOMP is the lowest, at 0.20%; its expense ratio is about 0.86% lower than that of the Morningstar category. The expense ratios for NVQ, QRFT, AMOM, and AIEQ are the same, 0.75%, and are way higher than the other two AI-powered ETFs. This may occur because these four AI ETFs utilize more advanced AI technologies and models (IBM Watson & Qraft AI-Enhanced Technologies) to screen and select stocks, thus imposing more costs on the fund and increasing expense ratios. Overall, all 6 AI-powered ETFs are comparatively less costly than the other funds in the Morningstar categories because they employ AI technology which significantly reduces costs compared to other ETFs.
Efficiency Measurement: Performances
The measurement of annualized return utilizes the NAV method. Where the changes in the ETFs’ net asset values become captured. To compare the returns of 6 AI-powered ETFs, the article chooses to use the 1-year annualized return. Because the inception date of NVQ is Dec 02, 2020, where only 1-year annualized return can become calculated. The performance section also includes the LOF return, the life of fund returns since the ETFs’ inception dates, and their 1-year annualized returns compared to those of the Morningstar categories and over the underlying indexes.
The graph illustrates that all 6 AI-powered ETFs generated negative annualized returns during the last year. However, their LOF returns are all positive, with NVQ of 13.29% the highest among them. This may happen because NVQ chooses to invest in value stocks adjusted for intangible assets, where it obtains the asset values of its portfolio more precisely, thus leading to a higher return when compared to other AI-powered ETFs. More importantly, all AI ETFs that employ Qraft AI-Enhanced Technologies generate more returns than those of the Morningstar categories. When comparing AI-powered ETFs to their underlying indexes, only AIVL and NVQ perform better, while the others underperform.
Efficiency Measurement: Risks
The risk measurement contains four factors: R-square, beta, alpha, and standard deviation. R-square explains how the movements of AI-powered ETFs are explained by their benchmarks; beta measures ETFs’ returns volatility to the whole market; alpha obtains the excessive returns of ETFs; standard deviation is calculated based on ETFs’ own returns volatility. All the factors use one year as the period for calculation.
As the graph above indicates, only AIVL and QRFT closely track their benchmark index, with an R-square value of 0.9831 and 0.9656. Consequently, their betas replicate the market beta, with 0.97 and 1. However, NVQ outcompetes other AI-powered ETFs because it only has a beta of 0.88, which is below the market. It is able to beat the market, thereby generating a 16.06% excess return over the benchmark.
Tradability Measurement: Trading Volumes & Bid/Ask Spread
For the tradability analysis, the article employs a 30-day median daily trading volume to estimate the liquidity and a 30-day median bid/ask spread to illustrate the transaction cost for investors.
According to the graph above, KOMP is traded in the market far more than the other AI-powered ETFs, with a trading volume 5 times more than that of AIEQ. NVQ’s performances are the best among the other ETFs, but its liquidity is too low, with a 30-day median trading volume of 782.25. The bid/ask spread for AIEQ is the highest at 0.33%. It is abnormal for a wide bid/ask spread to appear in a large growth ETF. However, this can happen because AIEQ is comparatively volatile. So market makers become eager for additional spread to compensate them for the risk that prices change.
Fit Measurement: Downside/Upside Capture Ratios
The Downside/Upside ratios indicate whether AI-powered ETFs outperform the market benchmark during market strength and weakness periods. An upside capture ratio over 100 shows that the ETF obtains more returns during periods of positive returns for the benchmark; a downside capture ratio less than 100 represents the ETF losing more than the benchmark when the benchmark is in the red.
The graph demonstrates that AI-powered ETFs can avoid loss during the benchmark downturn. However, AI-powered ETFs perform less effectively in the upside condition. AIEQ, KOMP, AMOM, and QRFT’s downside capture ratios are above 100, some even above 120, but their upside capture ratios are only 82.03, 77.34, 106.94, and 96.36, respectively. This may be due to the artificial intelligence utilized during the stock selection and the following portfolio adjustment processes that favor risks over returns.
Fit Measurement: Sharpe & Information Ratios
There are two differences between the Sharpe ratio and the information ratio. The first difference is the denominator, where the information ratio utilizes the tracking error of the tracking index as the denominator; the other difference is the numerator. Where the benchmark return becomes subtracted instead. The information ratio checks for the consistency of the investments in AI-powered ETFs on a risk-adjusted basis.
The above graph demonstrates the performances of AI-powered ETFs relative to the benchmark index. AIVL and NVQ generate exceptional returns, as their information ratios are 1.49 and 1.62. However, their Sharpe ratios are negative, suggesting a contradicting perspective. This can be possible if the portfolio return is higher than the benchmark return and lower than the risk-free return. For the rest of the AI-powered ETFs, they underperform the benchmark index.
To conclude, AI-powered ETFs still have a long way to evolve. They need to keep updating their AI technologies, adjusting their investment strategies, and rebalancing their portfolios to accommodate both the gain and loss situations in the market.
