Interactive Tool
Diversification: the only free lunch
Mixing assets that don't move in lockstep lowers a portfolio's risk without lowering its expected return: the closest thing investing has to a free lunch. It's free because of which risk it removes. Markets pay you for risk nobody can escape — the whole market falling at once — but pay nothing for the risk that one company stumbles, because anyone can delete that by owning many companies. Diversifying throws away only the unpaid kind, which is why the return survives the risk reduction. Meet it in five views: first the pure idea, where out-of-phase waves cancel; then the catch, where correlations aren't constant — they spike in the extremes and the free lunch shrinks just when moves are largest; then the messy reality of real, noisy returns; then its opposite, where a "500-stock" index turns out to ride on a handful of giants; and last, what actually decides each holding's weight.
Out-of-phase waves cancel 30% of the volatility. That shrinkage is diversification: exactly the portfolio's volatility falling below the average of its parts.
On the efficient frontier
Your portfolio's volatility above is its risk here: its spot on the horizontal axis. As you lower correlation, the frontier bows out and your mix slides left. (Return uses each asset's typical figure.)
An idealized model: perfectly smooth, repeating waves. Real returns are noisy and never cancel this cleanly. That's Part 2. Correlation sets each wave's phase; amplitude is its volatility.
Waves: the pure idea
- Each asset is a wave. Taller = more volatile; its correlation sets its phase. At +1 the waves line up; at −1 they're mirror images.
- The portfolio is their sum. The dashed lines show how big the volatility would be if risk simply stacked; the gap down to the solid wave is the risk that cancelled away. Slide correlation toward −1 and the portfolio flattens toward a straight line: near-perfect cancellation.
Shifting correlations: the catch
- Correlation isn't a constant. The single ρ in the variance formula is an average. In the extremes, assets that normally drift apart lurch into lockstep — correlation jumps toward +1 and the waves stop cancelling.
- It's not inherently good or bad. Watch the portfolio wave, flat in calm markets, swell to full size inside the shaded episodes — the joint move can be a rally or a sell-off, so the shading stays neutral. The statistics are identical; only the down move hurts. Diversification fades exactly when moves are largest, which is why the tidy average oversells it.
Messy reality
- Nothing cancels perfectly. With real, noisy returns the theoretical benefit and what a given run actually delivers rarely match, and there's no truly risk-free blend of risky assets.
- Assets sink together. The higher the correlation, the more often every holding is down at once: exactly when diversification helps least. That's why correlations, not just volatilities, drive portfolio risk.
Concentration: diversification in reverse
- "500 stocks" can mean a handful. A cap-weighted index hands the biggest companies the biggest share, so when a few giants swell, the top 10 names can make up a third or more of the whole. The effective number of stocks (1 ÷ the sum of squared weights) shows how few names you're really diversified across.
- Concentration is a bet. Equal-weighting every stock and cap-weighting the same 500 give very different rides: cap weight wins when the giants lead, and lags when they don't. Owning the index is convenient, but it isn't the same as owning 500 things equally.
This is the core idea behind Modern Portfolio Theory, introduced by Harry Markowitz in 1952: because correlations, not just volatilities, drive a portfolio's risk, the mix of assets can be engineered for a better risk-and-return trade-off. Formalized, that search is called mean-variance optimization: using expected returns, variances, and correlations to find the least-risk portfolio for a given expected return. Both views here are simplified, and real returns have fat tails the model can't capture. To see that optimization drawn as the efficient frontier, with a full range of outcomes, continue to Portfolio, Allocation & Bonds. Educational only, not financial advice.
Sources & further reading
How this tool was made
Mostly deliberate simulation: the waves, shifting-correlation, and messy-reality tabs generate synthetic return series so the mechanism is visible without real-world noise, and the index-weighting tab is an invented five-stock market. Only “How top-heavy is the market?” uses real data — S&P 500 constituent weights and concentration derived from CRSP records, 1928–present.
Research
- Markowitz, H. (1952). “Portfolio Selection.” The Journal of Finance 7(1): 77–91. The founding paper of modern portfolio theory.
- Samuelson, P. A. (1965). “Proof That Properly Anticipated Prices Fluctuate Randomly.” Industrial Management Review 6(2): 41–49.
Data
- CRSP S&P 500 Indexes & Daily Constituents, Center for Research in Security Prices, LLC, via WRDS. Source: CRSP®, Center for Research in Security Prices, The University of Chicago. Used with permission. All rights reserved. We ship only index-level aggregates — month-end concentration, index returns, and decade snapshots of the top-10 tickers by weight; no per-stock time series is redistributed. Pending written confirmation of the licence's scope for public educational use.
Educational use only, not financial advice. Every figure traces back to the sources above or to the inputs you set — the full method and the source code are public.