Risk & Money Management · Intermediate · 5 min
Diversification and correlation
You can explain why diversification reduces risk only when holdings are not strongly correlated, and spot hidden concentration in a portfolio.
Five shares, one bet?
Holding five different shares feels safer than holding one. If one company has a bad day, the others are still there.
But what if all five are software companies, and a single piece of news about the sector sends all of them down at once? Then the money was spread over five names, but not over five different risks. This lesson is about the difference.
Spread what moves differently
Diversification means spreading money across holdings so that a loss in one is not a loss in all. The SEC's investor guide describes it simply as not putting all your eggs in one basket. Economist Harry Markowitz developed the theory behind it in the 1950s, showing how combining assets that differ in return and risk can reduce the risk of the whole. He received the Nobel Prize in Economics for it in 1990.
The key ingredient is correlation, introduced in the forex chapter: a measure from −1 to +1 of how closely two prices have moved together in the past.
- High positive correlation, close to +1: the holdings rise and fall together. Adding more of them adds little protection.
- Low correlation, near 0: their moves are largely unrelated, so their swings partly offset each other.
- Negative correlation: one tends to rise when the other falls. That offsets risk most.
Hidden concentration is common:
- Same sector or theme. Several technology shares, or several funds that hold the same large companies.
- Same driver. Different currency pairs that all contain the US dollar.
- Same employer or country. Savings, job and home all tied to one company or one economy.
Two limits matter. First, diversification reduces the risk of individual holdings, but not the risk of the whole market falling. Second, correlations are measured over past periods and can change, sometimes quickly when conditions change.
Three trades, one idea
A fictional trader opens three forex trades, each risking 1% of the account: EUR/USD, GBP/USD and USD/CHF in the direction of a weaker dollar.
- On paper, that is three separate trades with 1% risk each.
- In practice, all three depend on the same thing: the US dollar. These pairs have often shown high correlation, positive or negative depending on how they are quoted.
- A surprise US interest rate decision strengthens the dollar sharply. All three stops are hit within an hour.
The trader lost 3%, on one idea. A rule that counts correlated trades together, for example a maximum of 2% total risk on any one currency, would have limited the loss.
Common mistakes
- Counting names instead of risks. Ten shares in one sector are closer to one bet than to ten.
- Owning several funds that hold the same companies. Check what is inside them.
- Adding risks per trade without checking correlation. Correlated trades can all fail together.
- Assuming correlations are fixed. They describe the past and can shift.
- Diversifying into things you do not understand. Every new holding brings risks of its own.
The takeaway
Diversification spreads risk across holdings that don't move together. Five assets that rise and fall together are one bet, not five.
Check yourself
Quick quiz
4 questions. Instant feedback, nothing is saved.
-
1Tap the portfolio that is really one bet.
Tap the chart.
Shares from the same sector react to the same news and conditions. Their high correlation makes five positions behave like one.
-
2When does adding a holding reduce portfolio risk most?
Holdings that do not move together can offset each other's swings. Highly correlated ones mostly add to the same risk.
-
3Three open trades, each risking 1%, are all on highly correlated currency pairs. What is the realistic risk?
If the pairs move together, all three stops can be hit by the same move.
-
4Why should correlation figures be read with care?
Correlation is measured over a past period. Relationships between markets can shift when conditions change.
of 4 correct.