White-label issuer Alpha Architect has struck again, with fund sponsor client Euclidean Technologies and their first foray into the ETF space, the Euclidean Fundamental Value ETF (ECML) .
Euclidean bills itself as a firm focused on "machine learning for long-term investing" and I like that the firm calls what it does machine learning, because it tells me that they are practitioners of what I'll call this scientific artform. Let's take a look at the firm and its new fund.
Euclidean: Machining the Fund
According to regulatory filings as of March 21, specifically form ADV Part 1 and Part 2, Euclidean was launched in October of 2011, although it's been running money since 2008. The company has two clients, with $111 million in assets under management. Fees range from 35 basis points (bps) to 150 bps and comes with an additional performance fee of 15% to 20% of returns above and beyond a dividend reinvested version of the S&P 500 index. Besides collecting fees, the folks at Euclidean have a mechanism set up that makes donations to "various charities focused on cancer-related research," which is explained in this shareholder letter. That the firm has been in business for the past dozen years and has been focused on data science and machine learning tells me they have had time to develop and hone their models with real-time data and not just back tested results. The latest edition of the deep learning models they employ was put in place at the end of March 2020, and the lack of flashing neon "AI" signs coupled with their longevity are good signs. I also like their non-traditional investment manager origin story.
ECML Targets the 'Undervalued, Underappreciated'
The fund sports a 95 basis point expense ratio, so $1,000 invested over a calendar year would be reduced by $9.50 due to fees over that period.
The overall strategy described in the summary prospectus is achieved by "investing in U.S. equities that (the company) believes are under-valued and under-appreciated by the market."
What is different about Euclidean's approach isn't so much what it's doing but how it's doing it. This starts with the universe of U.S.-listed equities and filtering out every company with market capitalizations below $1 billion. It has no explicit liquidity requirements, but my guess is it's counting on the larger names having more liquidity. They go on to backfill the model with historical financials, stock prices, and other data which is used to forecast a discounted earnings figure. They then calculate what they call an earnings yield by dividing the discounted earnings forecast by the company's total enterprise value. This is done for all names in the universe. There are subsequent steps that look to weed out so-called "Value Traps", or those names that have depressed valuations for good reason.
The final step involves taking the final model output of eligible portfolio names and returning the review process back to humans. Names are screened against data that the model "does not have access to" like press releases, bankruptcy filings, and general news, including corporate action announcements.
Looking through the holdings for ECML, I have to say that I am relieved. As I mentioned earlier, I like that Euclidean describes their process as "machine learning" and not "AI" and this security basket reinforces that position. What I like about this portfolio is the lack of those names that everyone who claims to be leveraging AI puts in their funds, like Apple (AAPL) , Tesla (TSLA) , Nvidia (NVDA) , Amazon (AMZN) , and Meta Platforms (META) to name a few. The fund is clearly making a bet on the consumer via housing and recreation names and some health care plays. There is also some exposure to energy and industrials. The names in question, though, are companies you've heard of but don't dominate headlines like the top five group of Preformed Line Products (PLPC) , Mueller Industries (MLI) , Mastercraft Boat Holdings (MCFT) , Boise Cascade (BCC) , and Builders Firstsource (BLDR) , which account in aggregate for 9.24% of the 114 names in the portfolio. I noticed Herbalife (HLF) is in the portfolio, too, and I'm sure they would say, "Hey, the model wants what the model wants."
Wrap It Up
The same conflict of interest disclaimers that have been mentioned in earlier articles appear in the fund's Statement of Additional Information (SAI), but there is also language that describes the steps the company takes to ensure fair distribution of trades across all the accounts they manage. All in, there are a few things I like about fund sponsor Euclidean Technologies and also the overall approach. One other plus is that the fund does not try to replicate S&P 500 sector weights, so it's not trying to be some kind of optimized S&P 500 index fund, but a pure alpha-seeking product. With a lower market capitalization bound of $1 billion, ECML works as an all-cap equity exposure vehicle as well. There are some links above that will take you to some historical returns, which since the implementation of the latest machine learning model in March 2020 seem to be outperforming the market. This is one fund that will definitely go on the watch list.
(AAPL and AMZN are among the holdings in the Action Alerts PLUS member club. Want to be alerted before AAP buys or sells these stocks? Learn more now.)
