Correlation and Concentration

Two holdings can hide the risk of one — check how closely they move together, not just how many you own.

• Explain what a correlation figure between two holdings actually measures, and read where a given number falls on the together-to-independent spectrum.
• Determine whether a given pair of holdings is genuinely diversified or functionally concentrated, based on their correlation figure rather than the count of holdings.
• Explain why a correlation figure describes past price behavior and should not be treated as a guarantee of future co-movement.

You already know the rule "don't put all your eggs in one basket." But here's the problem: two baskets can still hold the same egg twice, without you noticing. Owning two, five, or ten different stocks feels safe just because the names are different — but that feeling can be false.

The number that tells you whether your baskets are actually different is called correlation. Correlation measures how closely two holdings' prices move together, expressed as a number roughly between -1 and +1. A correlation near +1 means the two holdings tend to rise and fall together, almost in lockstep. A correlation near 0 means they move mostly independently of each other — one going up tells you almost nothing about what the other is doing. A negative correlation means they tend to move in opposite directions.

This is why counting holdings tells you nothing about real diversification. Diversification isn't about how many different names you own — it's about whether those names actually behave differently from each other. Two holdings with high correlation can behave like one big position wearing two name tags. Two holdings with low correlation genuinely spread your risk, because a bad day for one is unlikely to also be a bad day for the other.

The count says "two and two." The correlation number says something very different.

Look at two portfolios, same size, same number of positions. Portfolio A holds two IT-sector stocks with a correlation of +0.85. Portfolio B holds one IT stock and one FMCG stock with a correlation of +0.15. By position count, both look equally diversified — two holdings each. But they are not carrying the same real risk.

The count says "two and two." The correlation number says something very different.

In Portfolio A, because the two stocks move together 0.85 of the time, a single bad day for the IT sector hits both positions at once — that's functionally closer to holding one large position than two independent ones. In Portfolio B, the low 0.15 correlation means a bad day for the IT stock is far less likely to also be a bad day for the FMCG stock. Same count, very different real concentration risk. That gap is exactly what correlation reveals and position-counting hides.

Take two portfolios of the same total size, each holding exactly two stocks.

Portfolio A holds two IT-sector stocks. Their correlation is +0.85 — close to +1, meaning they tend to rise and fall together almost in lockstep. Portfolio B holds one IT-sector stock and one FMCG-sector stock. Their correlation is +0.15 — close to 0, meaning they move mostly independently of each other.

Now imagine a bad day for the IT sector — some negative news hits IT companies broadly. In Portfolio A, both holdings are IT stocks, and their 0.85 correlation means they're very likely to fall together on that news. The investor doesn't lose money on "one out of two" positions — they effectively lose on both at once, because the two positions were never really behaving independently. Two names, but functionally one bet.

In Portfolio B, the same bad IT-sector day hits the IT holding. But the FMCG holding, with its low 0.15 correlation to the IT stock, is far less likely to move the same way — FMCG demand and IT sector news generally have little to do with each other. So while one position in Portfolio B may dip, the other is not dragged down by the same cause. The overall portfolio absorbs the shock better.

Both portfolios had exactly two positions and the same total size. Neither looked "concentrated" by count. But Portfolio A was carrying far more real concentration risk than Portfolio B — and the only way to see that difference was to check the correlation figure, not the number of holdings.

Point: Correlation, not the count of holdings, determines whether a portfolio is genuinely diversified or functionally concentrated — a high correlation makes two positions behave like one, a low correlation means they carry independent risk.

One thing worth flagging before you rely on any correlation figure: it's calculated from past price behavior. A correlation of +0.85 means those two stocks moved together closely under the market conditions that already happened — it's a historical observation, not a promise. Correlation between the same two stocks can shift over time, sometimes sharply during market stress, when many things that used to move independently suddenly start moving together. That deeper behavior — how correlations shift across many holdings at once — belongs to later units. For now, treat any given correlation figure as a useful diagnostic check on the past, not a guarantee of the future.

• Believing that owning two (or more) different stocks automatically means being diversified. — Learners have been trained on the qualitative rule 'don't put all eggs in one basket' and judge diversification purely by counting distinct names or sectors held. Fix: Before trusting that a portfolio is diversified, check the correlation between its holdings. If it's high, two names may still be one real bet.
• Treating a correlation figure as a guarantee of how two assets will move together in the future. — The number feels precise and factual, so it's tempting to treat it as a fixed law rather than a historical observation. Fix: Remember correlation is calculated from past price movements under the conditions that produced it — use it as a useful check, not a promise about tomorrow.
• Assuming a high correlation number is always bad and a low or negative one is always good, no matter the context. — It's tempting to look for a simple 'good number vs bad number' rule instead of connecting the figure back to the actual goal of checking real concentration risk. Fix: Ask what you're trying to achieve. If your goal is genuine risk-spreading, low correlation supports it and high correlation undermines it — but the number itself is a diagnostic tool, not a target to chase for its own sake.

• The number of holdings you own is not the same as the number of independent bets you're actually taking.
• Correlation is a number, roughly -1 to +1, that shows how closely two holdings' prices move together — near +1 means together, near 0 means independent, negative means opposite.
• A high correlation (like +0.85) can make two holdings behave like one large position during a bad day.
• A low correlation (like +0.15) means one holding's bad day is unlikely to also be the other's bad day — that's real diversification.
• Correlation is calculated from past price behavior, so it's a useful check on the past, not a guarantee of future co-movement.

Q1. What does a correlation figure between two holdings actually measure?

• How closely the two holdings' prices move together, on a scale roughly from -1 to +1
• How much total money is invested in each holding
• How many years each holding has existed in the market
• The average yearly return of each holding Answer: How closely the two holdings' prices move together, on a scale roughly from -1 to +1 — Correlation is a number, roughly between -1 and +1, that shows how closely two holdings' prices move together — it says nothing about return, size, or age of the holdings.

Q2. A pair of stocks has a correlation of +0.90. What does this number suggest about how they move?

• They tend to rise and fall together almost in lockstep
• They move in completely opposite directions
• They move almost entirely independently of each other
• The correlation figure cannot tell you anything about their movement Answer: They tend to rise and fall together almost in lockstep — A correlation close to +1 means the two holdings' prices tend to move together closely — near 0 would mean independent, and negative would mean opposite.

Q3. True or False: If an investor owns three different stocks from three different companies, they are automatically well-diversified, regardless of correlation. Answer: False — The number of holdings doesn't determine real diversification. If those three stocks are highly correlated, they can behave like one large position on a bad day, even though they are three separate names.

Q4. A learner is comparing two new 2-stock portfolios of the same size. Portfolio X has a correlation of +0.80 between its holdings. Portfolio Y has a correlation of +0.10. Based on correlation alone, which portfolio is carrying more real concentration risk, and why?

• Portfolio X, because its holdings move together closely and a bad day for one is likely to be a bad day for both
• Portfolio Y, because a lower correlation number always signals higher risk
• Both carry identical risk, since each portfolio holds exactly two positions
• Neither carries meaningful risk, since both are already diversified by having two holdings Answer: Portfolio X, because its holdings move together closely and a bad day for one is likely to be a bad day for both — Portfolio X's +0.80 correlation means its two holdings tend to move together, so a bad day for one is very likely to also hit the other — functionally more like one position. Portfolio Y's low +0.10 correlation means its holdings behave far more independently, giving it genuinely lower concentration risk despite having the same number of positions.

Q5. Two stocks show +0.9 correlation over the past two years. An investor concludes: "These two will always move together at this same strength going forward." Does a strong past correlation guarantee the same future co-movement? Reveal: Weak: yes, +0.9 for two years is a reliable pattern to count on. Strong: correlation is calculated from past price behavior under the conditions that produced it — it describes history, not a guarantee; the conditions that caused that co-movement can change, and the correlation can shift with them.