Diversification is not a mood. It is a measurement, and the measurement has a shape: risk falls fast for the first few dozen names, then the curve goes flat. Everything you add after that point costs attention and edge while buying almost no additional safety.
Start with the most uncomfortable fact in equity investing. Since 1926, roughly 4 percent of all listed stocks account for the entire net wealth the stock market has created. The other 96 percent, taken together, did no better than short term government paper. This is not a quirk of one bad decade. It is the permanent structure of the market. Returns are not spread across the field like grass. They are stacked in a pyramid, and a thin capstone carries everything.
The same skew shows up inside company economics: only about 30 percent of firms earn returns above their cost of capital in any given stretch. Most stocks are, over their full lives, mediocre or worse. A handful compound for decades. Every portfolio decision you make lives downstream of this shape.
Now measure what adding names does. A one stock portfolio carries brutal single name risk: one accounting scandal, one lost customer, one regulator, and your capital is gone. Move from one name to five and the volatility of the whole drops sharply. Move from five to fifteen and it drops again. Somewhere between twenty and thirty names, the diversifiable part of the risk is essentially gone. What remains is market risk, the part no amount of adding tickers can remove.
Past that point the curve is a floor. Name number forty reduces portfolio risk by a rounding error. Name number eighty reduces it by nothing you could detect in a lifetime of statements. The safety was all purchased by the first few dozen positions. The rest of the shopping was decoration.
Here is what an eighty name portfolio actually is: an index fund with extra steps. Its returns will track the market to within noise, because that is what broad baskets do. But you paid for it in the most expensive currency you have. Eighty names cannot be known. You will not read eighty annual reports, track eighty competitive positions, or notice the quarter when name sixty three quietly broke. So you hold businesses you do not understand, at index level results, while telling yourself a stock picking story. If you want the index, buy the index. It is cheaper, it is honest, and it is guaranteed to hold the 4 percent that matter.
The opposite error is more dangerous. Concentration feels like conviction, and conviction feels like evidence. It is not. The only thing that licenses a large position is a recorded, audited edge: a written history of calls in a defined lane, scored against what actually happened. Our own case library is blunt about where edge lives. Calls made on undervalued quality hit 73 percent of the time. Calls made on overvalued names hit 42 percent, worse than a coin. A trader who concentrates inside a lane where his record shows 73 percent is doing arithmetic. A trader who concentrates because he is sure is doing astrology with a brokerage account. If you have no written record, you have no measured edge, and a portfolio built on feelings should be diversified to the point of boredom, because boredom is all that protects it.
So the doctrine is a barbell, and both ends are load bearing. First, own the field. A broad core holding guarantees you can never miss the 4 percent, and it collects the market's long run engine of roughly 6.5 to 7 percent real per year that two centuries of data keep paying. Nobody identifies the capstone companies in advance reliably. Owning everything is the only certain way to own them.
Second, hunt with a capped book. Our active side runs 19+1 slots and not one more. The cap is the discipline: a new idea must evict an old one, which forces every position to defend its place. Candidates come off Project 50, a standing watchlist worked in batches, and every entrant lands in one of four pools with sizing to match: A for core compounders, B for mature growth, C for momentum, D for moonshots. Geography is fixed at 75 percent USA, 25 percent India, so allocation is never a mood either. The field pays the base rate. The capped book is where a recorded edge, and only a recorded edge, is allowed to compound.