Direct answer: Position sizing answers the question 'how much of my portfolio should I put in this investment?' The answer depends on three factors: the investment's expected return, the probability that the thesis is wrong, and the correlation of the investment with your existing portfolio. Practical frameworks: for most individual investors holding 20 to 30 positions, a 3% to 5% position in any single stock is a reasonable default; above 10% in any single position requires exceptional conviction and should be acknowledged as a concentrated bet.
Research Protocol: Sizing a Position
Key Takeaways
- The maximum loss perspective: size any position so that if it goes to zero, the portfolio loss is acceptable. A 5% position that goes to zero costs 5% of the portfolio; an investor who cannot emotionally or financially tolerate losing 5% on a single position should size it smaller.
- Kelly criterion (theoretical optimal sizing): position size = edge / odds, where edge is the expected return per dollar risked and odds is the ratio of the potential gain to the potential loss. Full Kelly is too aggressive for most investors because it maximizes expected log wealth at the cost of large drawdowns; half-Kelly (50% of the Kelly-optimal size) is a practical heuristic that preserves most of the growth benefit while significantly reducing drawdown risk.
- Correlation matters: two positions that are highly correlated (both fall when interest rates rise, for example) provide less diversification than their individual position sizes imply. A 5% position in each of two identical-behaving stocks is a 10% correlated bet, not two independent 5% bets.
- Add in pieces, not all at once: initiating a position at 3% and adding to 5% if the thesis strengthens (and the price falls, making it more attractive) uses the investment process's evolving conviction rather than committing full size before the investment case is proven.
- The asymmetry of large losses: a 50% loss requires a 100% gain to recover; a 33% loss requires a 50% gain. Oversizing creates the mathematical asymmetry where large losses are harder to recover from than the percentage loss implies. Position sizing is therefore as much about avoiding catastrophic loss as about optimizing expected return.
The Equal Weight Starting Point
The simplest defensible position sizing approach is equal weight: allocate equally across N positions. For a 20-stock portfolio, equal weight is 5% per position. This is not optimal in the mathematical sense (it ignores differences in expected return and risk between positions) but it is robust because it avoids the problem of overconfident sizing in high-conviction positions that turn out to be wrong. Equal weighting forces diversification by construction and prevents the portfolio from being dominated by any single position. Academic research on equal-weighted portfolios finds they often outperform cap-weighted portfolios because they systematically over-weight small and value positions relative to cap-weighting, providing exposure to historically positive return factors.
Conviction-Scaled Sizing
If not equal weight, the alternative is sizing positions by conviction level. A practical three-tier system: full position (4% to 6%) for highest-conviction ideas where the investment case is thoroughly analyzed, the valuation is compelling, and the downside is clearly defined; half position (2% to 3%) for ideas where the case is solid but there is more uncertainty or the thesis is more speculative; starter position (0.5% to 1%) for ideas under active analysis where the decision is not yet final. This system prevents overcommitting before analysis is complete and allows position size to grow as conviction builds (adding to a starter position as more information confirms the thesis). The conviction tiers should be defined by the quality of the investment case, not by recent price performance.
Kelly Criterion in Practice
The Kelly formula: f* = (p times b - q) / b, where p is the probability of winning, q = 1 - p is the probability of losing, and b is the ratio of the win to the loss (e.g., if you win $2 for every $1 lost, b = 2). For an investment you believe has a 60% chance of returning 50% and a 40% chance of losing 25%: f* = (0.60 times 2 - 0.40) / 2 = (1.2 - 0.4) / 2 = 0.40. Kelly suggests 40% of the portfolio, which is aggressive. Half-Kelly suggests 20%; quarter-Kelly suggests 10%. These fractions assume your probability estimates are accurate; if they are overconfident (which research suggests they usually are), even half-Kelly is too aggressive. Most practitioners use the Kelly formula as an upper bound, not a prescription.
Avoiding Oversizing Mistakes
Common oversizing patterns: adding to a losing position without new positive information ('averaging down' due to the sunk cost fallacy and anchoring to purchase price); concentrated bets on a single macro theme (tech, crypto, or AI stocks all behaving similarly provide less diversification than they appear); employer stock overconcentration (employees with significant equity compensation are often overweighted in their employer's stock due to inertia and favorable tax treatment of restricted stock units); recent winner chasing (sizes increase in positions that have performed well, which is the opposite of value discipline). Position sizing discipline requires periodic review (not just the initial sizing decision) and a willingness to reduce positions that have grown to an uncomfortable percentage of the portfolio through appreciation.
Frequently Asked Questions
How many positions should I hold?
Research on diversification shows that most of the idiosyncratic risk (company-specific risk) is eliminated by 20 to 30 positions across uncorrelated sectors; beyond 30 positions, the marginal diversification benefit is small. More positions make the portfolio effectively index-like, which raises the question of whether active stock selection is worth the effort. Fewer than 10 positions exposes the portfolio to significant concentration risk from any single position's implosion. For most individual investors who want to be meaningfully diversified but can conduct thorough research on only a limited number of companies: 15 to 25 positions is a practical sweet spot.
Should I use the same sizing rules for index funds and individual stocks?
No. Index funds (total market ETFs, bond funds) can appropriately represent large fractions of a portfolio (40% to 80%) because they are diversified internally. The 3% to 5% individual stock position guideline applies to single-company risk; a 40% allocation to a total stock market ETF is not comparable to a 40% concentration in a single stock. A practical all-in portfolio: 70% in index funds (total U.S. market, international, bonds) as the core diversified base, with 30% allocated to individual stock ideas using 3% to 5% position sizing. This creates a core-satellite structure where the core provides baseline diversification and the satellite allows stock selection without catastrophic single-stock risk.
What if my position has grown to a large percentage through appreciation?
Position sizing applies at entry AND over the life of the position. A 5% position that grows to 12% (through price appreciation) is now a 12% concentrated bet regardless of the entry size. Whether to trim depends on: whether the investment thesis still supports the current valuation at the new higher price, the tax consequences of selling (embedded gain creates a real cost to rebalancing), and whether alternative uses of the capital are more attractive. A mechanical trim-to-target approach (selling back to 5% whenever a position exceeds 7% or 8%) imposes consistent discipline; a purely discretionary approach requires resisting the inclination to 'let winners ride' past rational position sizing limits.