Diversification That Actually Works

Diversification works exactly to the degree your holdings move DIFFERENTLY from each other. Twenty stocks that all rise and fall together provide the diversification of one stock, twenty times. The technical word is correlation; the plain question is: what single event would hurt all of these at once? If one headline nukes your whole account, you own one bet.

Different tickers, same bet: five tech companies that all breathe AI. A fund plus its own top holdings. Your employer's stock plus your salary plus your industry fund — one company failing hits all three. Diversification fraud is rarely intentional; it's what happens when you buy what you know, because what you know clusters.

The baskets that genuinely differ: sectors that respond to different forces, company sizes, geographies, and — biggest lever of all — asset types, since bonds and stocks often catch different weather. Our own model books are built as genuinely different fishing ponds on purpose, and we verify the overlap between them is near zero — because 'five strategies' that share the same names would be one strategy with five brochures.

Two truths to keep you humble: in a genuine panic, correlations lurch toward one — nearly everything falls together for a while, and only the most boring assets stand apart. And past correlation isn't a contract; relationships drift. Diversification is a seatbelt, not a force field: it makes the crash survivable, not impossible.

The math of diversification front-loads its benefits: going from one stock to a dozen-plus genuinely different ones removes most single-company risk; the hundredth position adds mostly bookkeeping. Past a point, more names just blur your attention and index-hug at higher cost. Diversify the RISKS, then stop. Collection is not construction.