AI-Powered Equity ETFs
The AIEQ ETF (AI Powered Equity ETF) trades on the NYSE (New York Stock Exchange). All the trading decisions are made by an AI computer algorithm.
As of August 2020, this ETF has been operating for 2 years and 9 months.
AIEQ Reviews in the News
Quite a few AIEQ review articles have been written. Surprisingly, most of the content is superficial. For example:
The article An AI-Powered ETF Failed Miserably at Beating the Market in 2018 — Here’s What You Can Learn From Its Mistakes by The Motley Fool, talks about the relative shortcomings of AIEQ.
“The S&P 500 index trounced the AIEQ, losing 6% compared to the ETF’s 2018 loss of nearly 16%,” comments the author, in order to point out the inadequacy of an AI-Powered ETF. After all, the S&P only lost 6%.
To which we say, a 6% loss is a victory???
In CNBC’s article, This ETF run by a robot is beating the market—here’s how it works, AIEQ is misrepresented as a stock. “[…] stocks including AIEQ collapsed in the fourth quarter of 2018 […]”
To which we say, AIEQ is an ETF, not a stock.
Even though both articles refer to the same time period, both AIEQ reviews miss the essence of what actually happened.
Shedding Light on AIEQ - How Similar to S&P 500 Is It Really?
To shed some true light on AIEQ, we collected some data from the Internet. Our analysis follows below.
Figure 1. We first plot the closes of the two series, and observe that they look a little too different. Ordinarily, we would expect that the two large-cap time series should look more similar. So, we need to find out what is happening.
Figure 2. The dividend record for AIEQ shows a large dividend payout of $1.903 on Monday, December 24, 2018 (on the day before Christmas). Thus, in Figure 1, the large decline on December 24 corresponds to a dividend payout to the investors – not a loss. Our analysis shows that both AIEQ review authors are mistaken (see above articles).
Next, when we see large changes in the cash flow structure, we need to adjust something to obtain a reasonable comparison. Sometimes, we might perform a custom cash flow analysis. It could be done in this case. However, we also can use the adjusted close for AIEQ, which is readily available, which will compensate for the price shock. Meanwhile, for the S&P index, we can use the close (because the close and the adjusted close are the same – no dividends).
Figure 3. So, next we compare the adjusted close of AIEQ and the close of the S&P. Now the two series look similar enough that we can compare the growth of the two series. For the life of AIEQ so far (2 years and 9 months) the growth of AIEQ is stronger than the S&P.
Figure 4. Next in our analysis, we make a cross-plot to compare the similarity of the two series. They are similar, but they are also different. How different? More in a moment.
(Recall that in one of our previous articles, we discuss how strongly the AIEQ appears to correlate with the S&P 500.)
Figure 5. Lastly, we compare two well-known large-cap indexes. The indexes are so similar that one series can be modeled in terms of the other.
Conclusion: AIEQ ETF Does Something Differently
Returning to Figure 4, we can conclude that the AIEQ algorithm is doing something different compared to the S&P index. This is also spelled out in the AIEQ fund management philosophy:
The system mimics a team of 1,000 research analysts
Investment Objective seeks long-term capital appreciation within risk constraints
Investment Process build[s] predictive financial models on approximately 6,000 U.S. companies, and derives an optimal risk adjusted portfolio consisting of companies with high opportunities for capital appreciation.
AIEQ constantly builds predictive models for approximately 6,000 companies (thus, they do not want to correlate with the S&P index). And, their stock selection process seeks high opportunities for capital appreciation (they are using an AI model and machine learning).
In principle, the more the algorithm learns, the less it will correlate with the S&P index.
Improving the AIEQ
Following our own AIEQ review, we see one important thing that the AIEQ algorithm still needs to learn.
When the broad market goes down, AIEQ should be moving up, not down.
If AIEQ performance can learn to go up, not down, that accomplishment will be a major breakthrough.