Bayesian Investor Blog

In 2015, I posted some investing advice for people who only spend a few hours per year on investing.

I intended to review it after five years, but a pandemic distracted me. It looks like this whole decade will end up being too busy for me to write everything that I want to write. But I’ve become able to write faster recently, maybe due to the feeling of urgency about AI transforming the world soon. So I’m getting a few old ideas for blog posts off of my to-do list, in order to be able to devote most of my attention to AI when the world becomes wild.

My advice worked poorly enough that I’m too discouraged to quantify the results.

I relied on two key factors that seem to have stopped working.

1. Stock market bubbles have been less important than I expected. That substantially reduced the value of my low-volatility investing strategy.
2. Some other changes that I didn’t expect have caused large-cap stocks to outperform small-cap stocks.

Pre-Pandemic

The 2015-2019 time period looked like a standard situation where I’d expect my strategies to work well. The mature stage of an economic expansion would normally have generated the kind of optimism that produces some bubbles. Yet I saw fewer than expected bubbles.

I guess I should conclude that there’s a general tendency of markets to become more efficient over time, as better information becomes available to investors.

Investors generally become more risk-tolerant as memories of the last crash fade, causing them to turn more attention to small-cap stocks. I didn’t see much of that pattern.

I get the impression that the most productive small companies have become less likely to be publicly traded. If so, that means that more than in the 20th century, a small cap stock EFT has been selected for stagnant or shrinking companies. Possible causes:

• Sarbanes-Oxley (2002) and the Dodd-Frank Act (2010) imposed new compliance and disclosure rules that require a larger fraction of a small public company’s resources than is the case for large companies. That deters small companies from becoming publicly traded.
• Equity funding has become more readily available to private companies, significantly reducing the need for good companies to go public. Investors in private equity have a bit more skill than the average investor at recognizing competent management, so a somewhat increasing fraction of companies that go public when they’re small are ones that private equity has rejected due to low competence.

The Unstable Decade, Part 1: Pandemic

The pandemic prompted some financial bubbles. There was a bubble in t-bonds – stock market advice wasn’t helpful there. There were bubbles in meme stocks such as GameStop and AMC – they didn’t have much effect on average investors.

There was a SPAC bubble. Avoiding volatile stocks was a good way to avoid losses here, but volatile SPACs didn’t get enough weight in the commonly used ETFs to make much difference.

There was a modest bubble in large cap stocks in the summer of 2020 that I called the nifty fifty. In hindsight, it looks like it was much less of a bubble than I thought at the time. It looked like a somewhat classic case of what my strategy was designed to avoid. Yet these stocks just weren’t as mispriced as were the nifty fifty of the 1970s, or the dot-com bubble.

The pandemic created big sources of volatility that were unrelated to whether a stock was subject to a bubble. So the valuable signals associated with low-volatility investing were drowned out by noise.

The pandemic also caused some shift in profits from small and medium sized retail companies to a few large companies, particularly Amazon.

The Unstable Decade, Part 2: The Prospect of AI Soon

There seems to be a broader trend of large companies performing well, both in terms of profits and stock price.

In the 20th century, there was a somewhat clear tendency for large companies to become stagnant. I attribute that largely to the problem of large, long-lived organizations developing interest groups within them, as described in Mancur Olson’s book The Rise and Decline of Nations. The process of those groups bargaining for rules takes decades to accumulate enough red tape to significantly harm an organization.

In 1960, nearly all top companies were 50+ years old. In 2025, half of the top 10 were 32 years old or less. 20 to 30 years less time to accumulate rot can make a significant difference in the extent to which a company becomes inflexible.

Economies of scale also seem to have become more important than was the case in the 20th century. NVIDIA, TSMC, and Amazon dominate their competitors more decisively than did IBM, Exxon, and General Motors.

Another important factor that is becoming increasingly important over the past couple of years is the highly uncertain effects of AI. There’s some rather unusual disagreement over the magnitude of AI’s impact 3 to 5 years from now. This creates volatility in important stocks (such as NVIDIA, Tesla) that is hard to distinguish from bubble-induced volatility. So far it looks like most of the evidence does not indicate a major bubble (I expect AI stocks will reach a bubble stage someday).

New Advice?

This is not a good decade for investors who hope to beat the market while only spending a few hours per year thinking about their investments.

The trends that will matter most are the ones that result from AI taking over increasing fractions of the work done by humans.

Markets are likely to continue to react with increasing efficiency to standard types of news such as earnings reports.

I used to focus heavily on understanding earnings reports, reading more of them than nearly all other investors, and identifying individual companies that were neglected. I’ve been moving steadily away from that focus over the past decade, devoting more attention toward evaluating industries and the technologies that drive them.

Markets have not yet figured out how to react to something as unusual as AI. I expect a few large reactions that are as hard to predict as were the reactions to the pandemic.

I’ve abandoned low volatility investing. I will likely retire from investing (due to AI making markets too efficient) without seeing conditions under which low volatility investing makes sense.

I still have a mild preference for small-cap stocks over large-cap stocks for now.

I expect a one-time boost to small-cap stocks later in this decade, when AIs take over most company-specific market research. Currently, small-cap stocks are often overlooked because competent analysts only have enough attention for in depth understanding of 20 or so companies. Such analysts end up focusing on the largest companies for which they can see room for improved understanding. AIs that are able to match human-level analysis of a given company will have the mental capacity to extend that analysis to all publicly traded companies.

After that one-time effect, I’m unsure how small-cap stocks will perform relative to large-caps. AI will enable small companies to more cheaply meet regulatory burdens. That effect might be offset by continued improvements in private equity markets. Any resulting difference in performance is likely to be small compared to the historical outperformance of small-cap stocks, as AI will eliminate the predictable driving force that drove that outperformance.

Fundamental investing will become harder to do well, as values will be changing faster than current reporting rules are designed to handle. Standard patterns of what constitute reasonable price/earnings ratios and dividend yield will likely become obsolete as economic growth rates undergo unusual changes.

I don’t know whether trend-following strategies will work.

I’m avoiding bonds, since AI may increase interest rates by accelerating economic growth. I expect bond yields to roughly match NGDP growth rates.

Understanding the rapidly changing impact of AI will likely be the most valuable thing you can do in order to invest well. If you can’t devote much time to that, then it’s likely futile to worry about whether you’re going to outperform the market.

Put some wealth into several differnt ETFs. In worlds where wealth is your biggest concern, you’ll end up doing better than most historical trends suggest. The amount that you put into stocks may end up being a good deal more important than any attempt to pick winners.

P.S. – Was the recent reaction to DeepSeek a CCP psyop? DeepSeek itself doesn’t look much like a psyop. But if I were running the CCP’s psyop team, the memes that triggered the January 27 panic in AI stocks look very much like the memes I’d want to spread in the US.

Money managers want to believe that AI will fizzle out before reaching a level that makes those managers obsolete, so they’re somewhat predisposed to imagine that DeepSeek’s efficiency will mean that we reach the AI fizzle stage sooner than expected. Of course that also means the money managers could have generated these memes themselves.