Reference

Five long positions in NVDA, AMD, AVGO, TSM, MU is one trade.

Position-sizing discipline at the per-trade level breaks down if your trades are correlated. Five 1 %-risk positions in semiconductors is effectively a 5 %-risk position in the semiconductor sector.

The correlation matrix problem

Single-trade position sizing assumes each trade is independent. In reality, equity positions are correlated — with the broader market, with their sector, with macro factors like rates and the dollar. A portfolio of 10 long positions all in US technology has implicit beta to NASDAQ-100 of approximately 1.0; on a 5 % NASDAQ-100 down day, the portfolio drops 5 %.

The position-size discipline budgets risk on the assumption that one position can lose without the others losing simultaneously. Correlated positions break this assumption.

Effective independent positions

For a portfolio of n positions with average pairwise correlation ρ, the effective number of independent positions is approximately n / (1 + (n − 1)ρ):

Number of positions	Avg correlation	Effective independent positions
5	0.0 (uncorrelated)	5.0
5	0.3 (mixed sectors)	2.5
5	0.5 (broad equity)	1.7
5	0.8 (same sector)	1.2
10	0.3	3.6
10	0.5	2.0
20	0.3	4.2
20	0.5	2.0

The diminishing returns to diversification are clear: at ρ = 0.5, you reach effectively 2 independent positions and stay there regardless of how many names you add. This is the empirical reason why portfolios of 30+ stocks rarely outperform a 15-stock portfolio of similarly correlated names — the additional names add cost without adding diversification.

Practical correlation buckets

Same name, different products. AAPL common stock + AAPL options ≈ correlation 0.95. Treat as one position.
Sector concentration. NVDA + AMD + AVGO ≈ pairwise correlation 0.7–0.85. Effectively a sector bet.
Long broad equity. S&P 500 stocks pairwise correlation typically 0.3–0.5 in normal markets, rising to 0.8+ in crisis periods.
Long stocks vs. long bonds. Long-run correlation ~0; recently positive in inflationary regimes (2022). Real diversification depends on the regime.
Long equity vs. long gold. Long-run correlation ~−0.1; a genuine diversifier across most regimes.

Adjusting position size for correlation

The simple rule: scale risk per position by the inverse of the effective-independent-positions count for your portfolio.

Worked example. You typically run 1 % per trade. You currently hold three semiconductor longs at average pairwise correlation 0.8. The effective independent positions = 3 / (1 + 2 × 0.8) = 1.15. Adding a fourth semiconductor at 1 % per-trade risk takes effective sector exposure above 4 %. To stay consistent with the original 1 %-per-position discipline, the fourth position should be sized at roughly 0.25 % — or substituted with an uncorrelated-sector long at the standard 1 % size.

The crisis-correlation trap

Pairwise equity correlations rise sharply during market stress. The 0.4 normal-market correlation between two utility stocks can spike to 0.85 in a March-2020-style flush, when investors sell broad-market index futures and the entire market moves down together. The position-sizing discipline that assumed normal-regime correlations breaks under stress — exactly when you need it most.

The mitigations:

Size positions assuming a stress-regime correlation, not the normal-regime correlation.
Hold a portion of capital outside long-equity positions (cash, treasuries, or genuinely-uncorrelated assets).
Use defined-risk options strategies (vertical spreads, butterflies) for high-conviction names instead of larger common-stock positions.

What the calculator does not handle

The position-size calculator on this site treats each trade independently. It does not aggregate correlation across your existing portfolio. For multi-position portfolio risk, you need a separate correlation-aware tool or a simple spreadsheet that tracks per-sector and per-factor exposure as a percentage of total capital.