Discover more from The Deload
AI: The Great Investor That Does Less
Teaching machines to sit on their ass
The Deload educates you about where the world is going to make smarter growth investments. If you haven’t subscribed, join more than 1,500 rational optimists:
Forgetting Charlie Munger
“The less you do, the more you do.”
An iconic line from a silly movie.
Paul Rudd plays a stereotypical surfer bro in Forgetting Sarah Marshall where he’s tasked to teach the main character, Jason Segel, to surf. To start the lesson, Rudd tells Segel, “Don’t do anything. Don’t try to surf. The less you do the more you do.” Rudd keeps telling Segel to do less until Segel just lays there on the board, literally doing nothing, to which Rudd says, “Well, you gotta do more than that.”
Great investors follow Rudd’s surfing mantra: The less you do, the more you do.
Charlie Munger advocates “Sit on Your Ass Investing.” There are only a few truly great companies that offer outstanding long-term performance. Those power law investments generate legendary returns. It doesn’t matter if your strategy looks for AAPLs that adhere to Buffett’s Margin of Safety with minimal downside and strong upside or early METAs that fit Thiel’s principle of monopoly where there’s a chance for a 100x+ gain.
Spend all your time finding those few great companies, buy them, then sit on your ass. Munger is probably a great surfer.
AI will prove to be a great investor, even better than Buffett and Munger, but AI needs to learn to sit on its ass to do it.
IBM Watson: Great at Jeopardy, Not at Stocks
Watson showed the world what AI could do long before ChatGPT.
The program started in the mid-2000s with a mission of beating humans at Jeopardy. After five years of development, Watson was regularly beating humans at the game show. Then it faced off against the best human Jeopardy player ever, Ken Jennings. It beat Jennings too, marking a major win for machine intelligence over humans.
On the heels of its triumph in Jeopardy, IBM expanded access to Watson to businesses looking to harness the power of machine intelligence, including investment funds.
Watson powers the actively managed AIEQ ETF, which was the first actively managed ETF powered by AI. ETFMG, the manager of AIEQ, describes the strategy as follows:
AIEQ applies AI technology to build predictive models on 6,000 U.S. companies. Each company has four underlying deep learning models: a Financial, News and Information, Management, and Macro model. Each of these four models have many underlying signals as depicted. The models identify approximately 30 to 200 companies with the greatest potential over the next twelve months for appreciation.
Despite using the powerful Watson, AIEQ has underperformed the S&P 500. Since its October 2017 inception, AIEQ is up 19.5% vs the S&P 500 up 67%. The story is the same YTD with AIEQ up 4.5% vs the S&P up 11.7%.
Eric Balanchunas from Bloomberg’s definitive Trillions ETF program thinks it’s because Watson trades too much.
In AIEQ’s fiscal 2021, the fund’s turnover rate was 540%. It was 1,700% in fiscal 2022. So Watson has the AIEQ trade a good amount. Perhaps that should be expected.
AIEQ relies on Watson to analyze “millions of data points across news, social media, industry and analyst reports, financial statements.” Pumping as much information as possible into AI is a common strategy because the strength of AI is thought to be its ability to process immense data. Massive data analysis makes for great marketing, but what you feed AI dictates how the program will act.
When you feed AI a ton of noisy data like news and social media, you’re setting it up to trade. All you can do with noise is trade it. You can’t invest in noise because it doesn’t matter for the long-term outlook of the actual company you’re investing in. A fundamental investor must ignore noise to be any good.
Balanchunas followed his overtrading assessment of AIEQ with a simple idea: Watson would probably be better off if it traded less and acted more like an investor. In other words, AI should sit on its ass more.
Of course, that would necessitate a wholesale change to the philosophy behind Watson’s AIEQ strategy, including the types of data it’s fed. More broadly, it would also necessitate a wholesale change in how investors think about using AI for investment strategy.
AI’s superpower relative to humans is thought to be its ability to process massive amounts of data to make more intelligent decisions. But great investments aren’t made by processing massive amounts of data. They’re made by focusing on a few important pieces of data, ignoring the noise, and then holding on while the compounding of a great business works its magic.
Ignoring the temptation to process as much data with AI as possible is a contrarian perspective that will yield better long-term investment results.
Intelligent Alpha: Sit On Your Ass AI Investing
My belief that AI will prove to be a great investor like Buffett and Munger drove me to build a series of AI-powered investment strategies. I now call them Intelligent Alpha (formerly known as Intelligent Indices if you’ve been following along. More on the name change at the end of this post).
Instead of trying to harness AI’s superpower to process massive amounts of data, Intelligent Alpha uses a different AI superpower: the lack of human emotion and bias that corrupts mere mortals from exercising the discipline of Buffett and Munger.
Intelligent Alpha’s investment committee of ChatGPT, Bard, and Claude is injected with an ownership mentality that avoids emotional mania buying or panic selling. I instruct the AI committee to seek long-term investments fueled by fundamental business performance, not just look for tickers that might go up because of some noise. Additionally, the portfolio review process is intentionally infrequent to reduce portfolio turnover.
The contrarian approach of avoiding the temptation to pump AI with excess data combined with the beautiful simplicity of looking for great companies yields a novel approach to AI-powered investing.
Since inception, Intelligent Alpha’s most comparable broad large cap strategies are beating Watson’s AIEQ trading approach:
Beautiful simplicity is hard to sell in the short term. It seems too easy, too obvious, but it often works in the long term. Time will judge whether Intelligent Alpha’s investment-focused is better than AIEQ’s trading focus, but I’m sure that simplifying the complexity invited by feeding AI too much data will remain an opportunity for alpha across industries well beyond investing.
Disclaimer: My views here do not constitute investment advice. They are for educational purposes only. My firm, Deepwater Asset Management, may hold positions in securities I write about. See our full disclaimer.
Announcement: Intelligent Indices is now Intelligent Alpha
What's smarter than smart beta?
I'm rebranding my Intelligent Indices project to Intelligent Alpha. The new name reflects the mission of the 30 investment strategies I've built so far: Beat legacy stock indexes with AI.
Why the name change?
Indexes aren’t built to beat other indexes. They’re built to represent certain segments for benchmarking. That’s not a useful application of the power of AI.
I expect AI will beat legacy indexes by eliminating the flaws of human emotions in investment processes, an appreciation for fundamental business performance that indexes definitionally ignore, and the consistency of a long-term investment mentality.
So far, so good. Many Intelligent Alpha strategies are beating legacy benchmarks by 200 bps or more since inception.
This is only a sampling of the many strategies AI can power.
Intelligent Alpha tracks a concentrated Intelligent Large Cap Conviction strategy that’s beating the S&P 500 by almost 200 bps and the Dow by about 50 since inception. The Intelligent Canada Select strategy is beating the TSX Composite by over 360 bps since inception. The Intelligent Japan Select strategy is beating the Nikkei 225 by 110 bps since inception.
It's still early, but the time for AI-powered investment strategies is now.
The future of investing is intelligent.