C.W.K.
Stream
Lesson 04 of 06 · published

Diversification — eggs in many baskets, mathematically

~30 min · diversification, portfolio

Level 0Numeracy Apprentice
0 XP0/68 lessons0/14 achievements
0/100 XP to next level100 XP to go0% complete

Eggs in many baskets — but the math is what makes it work

Everyone knows the saying. Don't put all your eggs in one basket. Spread your investments. But why does it actually work? What makes diversification more than folk wisdom?

The answer is correlation. From last lesson: when two stocks have correlation less than +1, combining them produces less volatility than either alone (weighted by holdings). The lower the correlation, the bigger the volatility reduction. With enough uncorrelated assets, you can shrink portfolio σ dramatically while keeping expected return roughly the same.

This is sometimes called the "only free lunch in finance." Most things in markets involve a tradeoff — higher return for higher risk, etc. Diversification is unusual: it lets you reduce risk without giving up much return, just by mixing things that don't all shake the same way.

The math intuition (no formula needed)

If you hold N stocks with equal weight, equal individual σ, and they're all perfectly uncorrelated (ρ = 0 between every pair), the portfolio's σ is roughly:

Same √t pattern from the last few lessons. Hold 4 uncorrelated stocks: portfolio σ is half the individual σ. Hold 25: portfolio σ is one-fifth. Hold 100: one-tenth. Magical, until you remember real stocks aren't uncorrelated.

In reality, most stocks within a single market are correlated to some degree (maybe ρ ≈ 0.3 between random pairs). So the diversification benefit hits diminishing returns — adding more US large-caps to your portfolio of US large-caps stops helping much after about 25-30 holdings. Beyond that, you're not really diversifying.

Diversification across what, exactly?

To keep the magic going, you need correlations to stay low. Real diversification happens across:

  • Sectors — tech vs. healthcare vs. utilities
  • Asset classes — stocks vs. bonds vs. real estate vs. commodities
  • Geographies — US vs. Europe vs. Asia vs. emerging markets
  • Styles — value vs. growth, large-cap vs. small-cap
  • Currencies (when investing internationally)

The cross-correlations between these tend to be lower than within-sector correlations. That's why a well-diversified portfolio holds a mix.

What diversification doesn't protect against

Diversification reduces idiosyncratic (stock-specific) risk. It doesn't reduce systematic (market-wide) risk. If the entire market drops 30%, your diversified portfolio of 50 stocks still drops something close to that. You can't diversify away "the whole economy is having a bad year."

This is why 2008 and 2020-March were so painful even for diversified investors — almost everything dropped together. Correlations spike to near 1 in crises ("when there's blood on the streets, all assets correlate"). Diversification helps in normal times, less in panics.

Track 9 distinguishes idiosyncratic from systematic risk formally. For now, the picture: diversification handles the noise; the market itself remains a risk you can't escape (only manage with hedges or holding cash).

The takeaway

Diversification works because correlations between assets are usually less than +1, and combining imperfectly-correlated returns reduces portfolio σ. Roughly: σ_portfolio ≈ σ_individual / √N for uncorrelated assets, with diminishing returns past 25-30 holdings within a single market. Cross-sector, cross-asset-class, cross-geography diversification keeps the magic going. It doesn't eliminate market-wide risk — but for the noise, it's the closest thing finance has to a free lunch.

Exercise

  1. You hold 1 stock with σ = 30%. Approximate σ if you held 9 uncorrelated stocks each with σ = 30%?
  2. Approximate σ if you held 100 of them?
  3. Now suppose those 100 stocks have ρ = 0.5 with each other (more realistic). Will the portfolio σ be closer to your answer in (2), or closer to 30%? Why?
  4. Why does diversification work less well in 2008-style crises?

Progress

Progress is local-only — sign in to sync across devices.
Spotted a bug or have feedback on this page?Report an Issue
💛 by Ttoriwarm

Comments 0

🔔 Reply notifications (sign in)
Sign inPlease sign in to comment.

No comments yet — be the first.