I do not believe in diversification.

“Do not put all of your eggs in one basket.” This is the cardinal rule taught in every Finance 101 class. The statistical mathematics behind it is elegant: each stock carries its own idiosyncratic risk, but a portfolio containing many stocks averages over these risks, leaving the investor exposed only to systematic market risk.

We see a parallel in statistics with Random Forests. A single decision tree often suffers from high variance, but an ensemble of trees averages out these variances, resulting in a model with robust predictions.

Similarly, in life, it often seems reasonable to engage in many different activities to hedge against the risk of failure. An adolescent aspiring to be a comedian or artist might be advised to focus on academics and major in business to hedge against the uncertainties of a creative career. By lining up a corporate safety net while pursuing creative endeavors on the side, the total risk is theoretically reduced. To diversify further, this student might also join the baseball team, hoping for an athletic scholarship to strengthen their position.

One can imagine a myriad of paths an adolescent might draw to “optimize” their life. However, so far, we have only discussed risk and loss, or “Value at Risk” (VaR). We must also consider the potential upside. With every focused endeavor, there is a non-zero chance of hitting it big. The adolescent might become a Managing Director in investment banking, a national celebrity comedian, or a professional athlete.

This has always been my strongest argument for engaging in new activities or meeting new people. If the downside is minimal but the potential upside is uncapped, it is only a question of time before I capture a “fat tail” event. This is the core lesson I took from Nassim Taleb: one should be antifragile against extreme negative outcomes, but positioned to gain from extreme positive ones.

From a stochastic point of view, these derivations make sense. In portfolio allocation, where we deal with a continuous scale (fractions of a portfolio) and continuous returns, the math is straightforward. However, this view changes when variability decreases, the choices become binary, and the rewards become deterministic.

Consider big, lumpy choices, like real estate. Do we invest one million dollars in Neighborhood A or Neighborhood B? It is not straightforward to buy a fraction of an apartment in A and a fraction in B. Even if you could (via REITs), it might be nonsensical if you possess local knowledge that Neighborhood A is superior.

A more pertinent example is the small business owner. Often, a large fraction of their net worth is tied up in their business, exposing them to massive idiosyncratic risk. Should we advise this owner to downsize a cash-flowing operation just to buy stocks and bonds for the sake of diversification? It would be foolish to starve a profitable business of capital when the owner could generate far greater returns by reinvesting in their own operations, for instance by increasing ad spend to generate leads.

As Charlie Munger noted in an interview with the Hoover Institute, diversification is often a strategy for the “know-nothing” investor. Similarly, Peter Thiel has argued that higher education is often just a substitute for thinking about the future, essentially a form of hedging. If one possesses strong conviction and deep knowledge, over-diversification (to the level of an ETF) becomes unnecessary.

Yet, this year, I was guilty of sticking to the “know-nothing” strategy, even when I knew better.

At the beginning of the year, Google was (and remains) a phenomenal company. It boasts world-class research and development, anchoring an ecosystem of services ranging from Waymo to YouTube. Earlier in the year, the market discounted the stock due to uncertainty regarding the company’s AI future and looming anti-trust litigation. In my view, these fears were unfounded; Google has always been a pioneer in Machine and Deep Learning. Consequently, compared to its tech peers, Google traded at a PE of only 20, which was a steal by historical standards.

I hesitated. I watched the stock trade at $150 in April; today, it sits above $300. Ironically, Berkshire Hathaway later announced a significant stake in the company, further bolstering the price. While I am not worse off financially, I would have been significantly wealthier had I listened to Munger. I erred on the side of inaction because I clung to the dogma that I could not outperform the market. While statistically true in the aggregate, in this specific instance I possessed a clear understanding of the risk-reward asymmetry. I chose to insure myself against potential ignorance, but the premium was exorbitant: while the S&P 500 returned approximately 16%, Alphabet stock surged 85% in the same six-month period.

As with asset pricing, the key question is pricing. Does the value this insurance creates align with the premium I am paying? A great company might have a distinct moat and cash flow, but its stock might still be too expensive. The same applies to diversification. In my Google example, the premium paid for diversification was too high.

We must ask the same question about any personal policy we opt into to reduce risk: does the cost of the hedge outweigh the potential loss? With financial options, the premium is explicit and tangible. With life choices, the cost is latent, hidden in the “what could have been.” Consider the adolescent again. All those layers of risk reduction, such as the business degree and the backup plans, may eventually cost him the very thing he sought: a career in the cultural scene. From this perspective, the four-year degree is not just an education; it is an incredibly expensive insurance policy paid for with the scarcest resource of all: time. As in any asset pricing problem, the question is not whether the insurance is “safe,” but whether the premium is simply too high to pay.