Correlation

Here is a question that catches almost every new investor, and the honest answer is not the obvious one. You own ten different stocks instead of one. Are you safer?

It feels certain that you are. Ten is more than one; spreading money across many names is the first piece of advice everybody hears. But the count of names is not the thing that keeps you safe. What keeps you safe is whether those names can fall at different times — or whether they all drop together the moment one shared worry arrives.

Imagine ten stocks that always move as one: when the market frowns, every one of them falls by roughly the same amount on the same day. You do not own ten holdings in any way that matters. You own one holding wearing ten different name tags. The brokerage statement looks spread out; the risk is not.

This module is about that gap — between diversification that is real and diversification that is only cosmetic. The single idea that separates the two is correlation: a plain measure of whether two holdings tend to move together, move apart, or move independently. Get correlation right and a small handful of positions can be genuinely safe. Get it wrong and fifty stocks can still be one bet.

The previous module weighed concentration against diversification — how much to hold in a few strong ideas versus how widely to spread. This module hands you the tool that decision actually needs. Because "how many stocks should I own?" is the wrong question. The right question is: how many different things can go wrong?

Diversification is often called the one free lunch in investing — the rare case where you can lower your risk without giving up your expected return, simply by holding things that do not all sink at once. But that lunch is only free if the holdings are genuinely different. Spreading across truly independent risks lowers how hard any single blow can land — and the word doing all the work in that sentence is independent. Ten holdings that share one driver give you none of the free lunch, because a blow to that driver is a blow to all ten at once.

Correlation is simply the name for that sharing. Two stocks are highly correlated when the things that move one also move the other — the same interest rate, the same monsoon, the same US technology budget, the same festival season. They are uncorrelated when the forces that push one have little to do with the forces that push the other. And in the worst case they are negatively correlated: when one falls the other tends to rise, so the two partly cancel out.

The trap is that correlation hides behind different company names, different logos, different founders. Five banks look like five companies. To the eye they are five. To the market, in a bad week for lending, they are close to one. The statement counts names; the drawdown counts drivers. This module teaches you to count drivers.

Correlation is a single number between −1 and +1 that answers one question: when one holding moves, what does the other tend to do?

• Near +1 — they move together. Up together, down together, roughly in step. Two large private banks in India sit here: not identical, but pulled by the same forces most of the time.
• Near 0 — they move independently. What happens to one tells you almost nothing about the other. A software exporter and a domestic cement maker might sit near here — different customers, different cycles.
• Near −1 — they move in opposite directions. When one falls the other tends to rise. True opposites are rare in stocks; this is why people reach for gold or bonds, which sometimes rise when equities fall.

You do not need to compute this number by hand, and chasing a precise decimal is not the point. The point is the instinct: before you add a holding, ask what already in my portfolio moves the same way as this? If the answer is "most of it," the new position is not spreading your risk — it is deepening a bet you already have.

The clearest way to feel it is a crash day. Put two portfolios side by side, each with five holdings, and drop the market. The first looks diversified — five separate stocks — but they are all lenders, so they all fall together. The second holds five things pulled by genuinely different drivers, so some fall hard, some barely move, and one even rises. The headcount is identical. The damage is not.

Look at the two totals. Same five holdings on each side, same terrible day for the market — and one book is down a third while the other has barely been scratched. Nothing about the number of stocks explains the difference. Everything about the correlation between them does.

Watch this happen in an ordinary portfolio. illustrative

A reader has ₹10,00,000 in the market and is proud of holding eight stocks — she has read that diversification matters and she has spread her money across eight names. The list: two large private banks, one small private bank, two non-bank lenders (an NBFC and a housing-finance company), a life insurer, a broking firm, and a payments company.

On paper, this is eight companies in eight boxes. Now ask the driver question: what makes each of these rise and fall? Interest rates. Credit growth. Bad-loan cycles. The health of the financial system. Every single one of the eight lives on those same few forces. When the market gets frightened about rising rates or a bad-loan scare, they do not fall one at a time in a polite queue — they fall together, on the same day, by similar amounts.

So her real position is not "eight stocks." It is one large, concentrated bet on Indian lending, chopped into eight slices. The eight names gave her the feeling of spread — the comforting sight of a long holdings list — while giving her almost none of the substance. This is naive diversification: adding names without adding drivers.

The repair is not "own more stocks." Twenty lenders would be worse, not better — the same bet, more finely sliced. The repair is to ask what else her money could ride on that does not move with lending: a consumer-staples business that sells soap and biscuits through every cycle, a pharma exporter earning in dollars, an infrastructure or utility name on a different clock, perhaps a slug of cash or gold that tends to hold up exactly when equities are falling. She does not need forty positions. She needs a handful of genuinely different ones.

The lesson generalises far beyond banks. Three IT names that all depend on US technology budgets and the rupee are one bet. A basket of small-caps that all soar and crash on the same risk-appetite tide is one bet. Even a spread that looks wonderfully varied in good times can quietly become one bet in a crisis — which is the sharpest edge of all, and the next thing to understand.

Correlation is a powerful lens, but it is a measured relationship, not a law of nature — and it has real limits you must respect.

It is measured from the past, and the past can lie about the future. A correlation number is calculated from how two holdings have moved. Nothing forces them to keep moving that way. Two businesses that drifted independently for years can suddenly lock together when a new shared risk appears — a policy change, a common lender, a shared customer.

Correlations rise toward one in a crash — exactly when you need them low. This is the cruel part. In calm markets, holdings across sectors move fairly independently, and your diversification looks solid. In a genuine panic — 2008, March 2020 — almost everything falls together as frightened investors sell whatever they can. The very spread you were counting on tends to fail on the one day it mattered most. Correlation is not a fixed shield; it thins precisely under the heaviest blow.

A low correlation is not a reason to own something bad. Correlation tells you how a holding fits, never whether it is worth owning. A weak company that happens to move independently is still a weak company. Diversification arranges good decisions; it cannot rescue poor ones.

The same handful of correlation mistakes catch beginner after beginner. Named once, they are easier to catch in yourself.

1. Counting names instead of drivers. "I own fifteen stocks" is a statement about a list, not about risk. If the fifteen ride on three shared forces, you own three bets. Always ask what would have to go wrong, and how many separate things that really is.

2. Mistaking a long sector list for spread. Five banks, or eight lenders, or a dozen small-caps that all track the same risk mood — the eye sees variety, the market sees one position. Different logos are not different drivers.

3. Trusting calm-market correlations to hold in a storm. The spread you measure in a quiet year is the best case. Assume it will weaken in a crisis, because it usually does, and size your positions for the day everything falls together rather than the day it does not.

4. Chasing more holdings as if the number itself were safety. Beyond a modest handful of genuinely different positions, adding names adds admin, not protection. Twenty correlated stocks are no safer than the two or three underlying bets they represent.

5. Forgetting the risks you did not choose. Your salary, your employer's stock, your home, your fixed deposits — these correlate with your portfolio too. If you work in a bank and own banks, one bad cycle can hit your job and your savings at once. Real diversification looks at the whole life, not just the demat account.