How many bets does a fund need?
Given the power law (Lesson 21), portfolio construction asks a genuinely counter-intuitive question: how many companies does a fund need to back to have a statistically reasonable chance of catching at least one outlier? Too few positions and the fund is essentially gambling on a single dice roll; too many and reserves (Lesson 22) get stretched too thin to matter for any of them.
Two schools of thought
| Strategy | Portfolio size | Reserve ratio | Logic |
|---|---|---|---|
| Concentrated | 15-30 companies | High (~1:1 or more) | Fewer, higher-conviction bets; defend ownership hard in each |
| Diversified | 60-150+ companies | Low (~0.3:1 or less) | Statistical exposure to more chances at an outlier |
Neither is objectively correct, the choice depends on a fund's ability to actually pick well. A team with genuinely superior selection skill benefits more from concentration, since their non-outlier picks likely still outperform average. A newer team, or one deliberately optimizing for broad market exposure, gets more value from diversification, since it doesn't depend on any single judgment call being right.
Batting average is not the goal
A fund with a 60% "win rate" (positive-return exits) but no true outlier will badly underperform a fund with a 20% win rate that caught one. This is the practical, portfolio-level consequence of Lesson 21: optimizing for fewer losses, rather than for a chance at the outlier, is optimizing for the wrong number entirely.
Checkpoint
- Portfolio construction: sizing the number of bets to balance outlier odds against reserve depth per company.
- Concentrated portfolios bet on selection skill; diversified portfolios bet on statistical exposure.
- Batting average across the portfolio is not the goal, catching the outlier is.
If anything here still feels unclear, ask before moving to Lesson 28.