Number of stocks in a portfolio

The concentrated versus diversified portfolio debate in M6.02 gave you the philosophy: your best idea far exceeds your tenth, so why dilute it (Munger, Buffett's 20-punch card, Keynes, Pabrai, Sleep); against which stands humility about your edge (Graham's baskets, Schloss's ~100 names, the index). This section turns that debate into a decision you can defend in writing, which is the module's gate.

What is the case for concentration?

The concentration case, in numbers. Diversification already did half the work: past ~10 names the risk benefit is nearly gone while the return dilution keeps mounting. The concentrator adds a second argument from knowledge: you can only truly know a handful of businesses at the depth this program demands (the 2–3 day deep dive is not something you can run on fifty names and keep current). Under a real edge, concentration is return-optimal. It is also the only honest position size for deep knowledge, because a 1% position is a confession that you don't know the business well enough to own more. Buffett's "diversification is protection against ignorance; it makes little sense if you know what you are doing" is exactly this: position size is a truth-teller about how much you actually know.

What are the three caps on concentration?

The three caps that concentration must respect (from M9.01 and M6.03, made concrete here):

1. The Kelly cap. Kelly sizes by edge, which is the mathematical engine of concentration. But full Kelly assumes you know the true odds, and you never do. Overestimating your edge (overconfidence, M6.04) turns full Kelly into a ruin machine, so serious concentrators use fractional Kelly (a half or a quarter), which already pushes you toward more, smaller positions than raw conviction suggests. Concentration bounded by fractional Kelly is disciplined; concentration bounded by nothing is gambling.
2. The ruin cap. A bet attractive "on average" can still ruin you on your one actual path through time if a single loss is large enough to knock you out (ergodicity, M6.03). So no single position, however high-conviction, may be sized so that its catastrophic case (fraud, a zero) permanently impairs the portfolio. This is a hard limit, not a preference: it is why even Buffett caps individual positions. A practical expression: if a name could plausibly go to zero (a lot can; accounting fraud catches even careful analysts), its position size times 100% loss must be a survivable dent, not a mortal wound.
3. The independent-bets cap. The number that matters is not names but independent bets. Ten names that are all Indian NBFCs, or all high-multiple US software, are one factor bet held ten times (diversification's ρ̄ is near 1 for them, so their systematic floor barely falls below a single name's; IL&FS/DHFL sank the whole NBFC complex together, M5.02). Conversely, ten genuinely uncorrelated businesses may carry less risk than thirty correlated ones. Concentrate your capital, but never your factor exposure by accident.

How do you choose your number?

The dial, and what sets it. There is no universal right number; there is a dial, and the master brief and M6.02 are explicit that reasonable masters sit at both ends. Set it toward concentration (say 8–15 positions) when: your edge is deep understanding you can actually maintain; the businesses are high-quality and durable (low per-name permanent-loss risk); your holdings are genuinely uncorrelated; and your temperament can hold through the violent drawdowns concentration guarantees without capitulating at the bottom. Set it toward diversification (25+ positions, or an index) when: your edge is statistical (deep value, quant, factor baskets where individuals fail but the group works); your information is unreliable (higher fraud/governance risk, a structural reason to diversify more, or diligence far deeper, in promoter-dominated Indian mid-caps); you cannot reliably rank your ideas; or you simply lack a demonstrable edge, in which case broad diversification / indexing is the honest, correct choice (see active vs passive investing). The old heuristic "concentrate to build wealth, diversify to preserve it" captures part of it, but the real determinant is always the quality and reliability of your edge set against the survival constraint.

What is a sensible default?

The synthesis most great investors converge on, and a reasonable default for a skilled analyst doing deep work on durable businesses, is roughly 8–20 genuinely independent positions: concentrated by index standards, diversified by Munger's, with each position sized by conviction × downside (fractional-Kelly logic) under a hard rule that no single name's worst case, and no single factor's turn, can sink the ship. You will commit your number and its reasons to writing in the mini-project; the argument for it is assembled here. The honest final word, per the program's guardrails: this is a genuinely contested question, and the right answer is the one you can execute under pressure given who you actually are. A policy you abandon in a 40% drawdown is worse than a more modest one you can hold.