NISM Series XXI-B · Chapter 15: Introduction to Modern Portfolio Theory

L1 · Foundation

Modern Portfolio Theory — the ground floor

Where Modern Portfolio Theory came from, what it assumes, how utility captures an investor's dislike of risk, and why correlation is the whole secret of diversification.

13 min read · Worth 3% of the paper · Written from pages 251–264 of the official NISM workbook · Updated 17 September 2026

Eggs, baskets, and a missing number

Every grandmother in India knows the rule: do not put all your eggs in one basket. Investors knew it too. Before 1950 they knew diversification helped. What they did not have was a way to measure how much it helped.

That changed in 1952. The Journal of Finance published an article called "Portfolio Selection" by Harry Markowitz. In 1990 he won the Nobel Prize in Economics for this "Portfolio Theory". His ideas now go by the name Modern Portfolio Theory (MPT).

How did Markowitz put a number on diversification? He used the statistical idea of covariance, or correlation, between assets. He showed mathematically that the variance of the rate of return is a meaningful measure of portfolio risk. Then he derived the formula for the variance of a whole portfolio.

The five assumptions

MPT rests on these assumptions. The exam likes them.

An investor wants to maximise return for a given level of risk. Given two assets with the same return, she picks the one with lower risk.
Investors see each investment as a probability distribution of expected returns over a holding period.
Investors maximise one-period expected utility. They give utility scores to the portfolio choices in front of them.
Investors base decisions solely on expected return and risk.
Investors estimate portfolio risk from the variability of expected returns of the assets inside it.

Risk, and the price of bearing it

Risk means more than one outcome is possible. Investors demand a risk premium for bearing it. Whether a given premium is enough is a central question in investment management.

The workbook's section heading names three kinds of investor — risk averse, risk seeking and risk neutral — but the text explains only the first. A risk-averse investor invests in risk-free options, or in options with a positive expected risk premium. The greater the risk, the greater the premium she demands.

Utility: one number for "how much do I like this?"

Investors give each portfolio a utility score. Higher expected return and lower risk earn a higher score. Higher risk and lower return earn a lower score.

The function the workbook uses is:

U = E(r) − ½ × A × σ²

U — utility value
E(r) — expected return
A — the investor's index of risk aversion
σ² — variance of returns

Read it in plain words. Return pushes utility up. Risk pulls it down. And how hard risk pulls depends on A — on this particular investor.

The decision rule: compare utility with the risk-free rate. If utility is greater than the risk-free rate, the investment is worth making. The workbook's logic is that utility is really a "risk-free equivalent" return for that investor.

Why utility is not the same as the Sharpe Ratio

The Sharpe Ratio and the risk-to-reward ratio also measure excess return against risk. But they leave out the investor's risk aversion. They assume every investor sees risk the same way. On them, a higher ratio simply means a better investment.

Real life is different. A good portfolio is not equally attractive to every investor. What explains that? Utility — because it builds in each person's unique risk aversion.

Ex-ante and ex-post

Ex-ante return — the expected return, found by multiplying each possible return by its probability and adding them up.
Ex-post return — a return calculated from historical data.

Ex-ante risk is the standard deviation of expected returns — the square root of variance. Variance measures how widely returns are dispersed around their expected value. Larger variance means larger risk.

From one security to a portfolio

Portfolio expected return is simple. It is the weighted average of the expected returns of the assets in it. Nothing surprising.

Portfolio risk is where the magic lives. It is not a simple weighted average. It depends on three things:

the weights of the investments
the risk of each investment
the co-movement between investments

That third item is covariance.

Covariance and correlation

Covariance measures how far two variables "move together" relative to their own averages.

The correlation coefficient is covariance standardised — divided by the product of the two standard deviations:

r(ij) = Cov(ij) ÷ (σi × σj)

Correlation runs from −1 to +1.

Correlation	What it means
+1	Perfect positive correlation — returns move together in a completely linear way
−1	Perfect negative correlation — returns move in opposite directions
Less than +1	Some diversification benefit exists

Counting the covariance terms

Every pair of assets adds a covariance term.

Two assets: three terms — two weighted variances and one weighted covariance.
Three assets: six terms — three weighted variances and three weighted covariances.

The formula for the number of covariance terms is (n² − n) ÷ 2.

A warning for the exam: the workbook gives the 50-security count twice, and differently. Section 15.7.2 says 1,125 weighted covariances. Section 15.11 says 1,225 correlation estimates, citing the same formula. Working the formula gives (2,500 − 50) ÷ 2 = 1,225. Know both figures exist.

What correlation does to the picture

Plot return on the y-axis and risk on the x-axis.

Correlation = +1: every mix of the two securities sits on a straight line. You can never beat the return of the higher-return security or the risk of the lower-risk one. No diversification benefit at all.
Correlation below +1: portfolio risk is less than the weighted average of the two risks. The curve bulges to the left. Some mixes have lower risk than either asset alone. This is the essence of diversification.
Correlation = −1: the complete benefit is realised. A particular mix could give a zero-risk portfolio.

The efficient frontier

Now combine many securities in many weights and plot them all. Some combinations give the highest return for a given risk, or the lowest risk for a given return. Together they form an umbrella-shaped curve: the efficient frontier.

A portfolio below the frontier is beaten. If portfolio C has the same return as A but more risk, A is efficient and C is not.

Optimisation, and the risk of bad inputs

An optimum portfolio combines investments with desirable risk–return traits for a given set of constraints. To use MPT, the manager must estimate three inputs:

the expected return of every asset in the universe
the standard deviation of each asset's expected returns
the correlation coefficients between every pair

Plus any constraints the investor has. From the feasible portfolios, pick the one that meets the investor's objectives.

The catch: the output is only as good as these statistical inputs. With 50 securities there are 1,225 correlations to estimate. The error that creeps in from all this estimating is called estimation risk.

Beyond the book

In India the same maths sits behind every multi-asset allocation a portfolio manager builds. Equity, government bonds and gold are popular combinations precisely because they have not always moved together. The lesson of MPT is that the mix, not any single holding, decides portfolio risk.

Think of it like this

Correlation is two umbrella sellers and two ice-cream sellers

Picture a stall on Marine Drive. If you own two ice-cream carts, a rainy week hurts both at once. Their earnings are highly correlated — owning two carts does not make your income steadier.

Now own one ice-cream cart and one umbrella stall. Rain kills ice-cream sales but sells umbrellas. Sunshine does the reverse. The two incomes move in opposite directions, so your total is much smoother than either stall alone.

That is the mechanism of MPT. Risk falls not because you own more things, but because what you own does not move together. At a perfect −1, the workbook says a zero-risk mix is possible.

Utility is the same thali judged by two diners

A spicy Andhra thali arrives. One diner loves heat; another cannot handle it. The thali is identical — same food, same price. Yet one rates it 9 and the other 4.

The Sharpe Ratio is like a restaurant rating: one number for everyone. Utility is each diner's own score. The spice is the risk, the food is the return, and the diner's tolerance is A, the risk-aversion index. A high-A investor gets a much bigger penalty for the same risk — which is why one good portfolio is not equally attractive to every investor.

The efficient frontier is the best seats in a cinema

In a cinema hall, several seats cost the same. Some have a clear view; some sit behind a pillar. No sensible person pays the same price for the pillar seat.

The efficient frontier is the row of clear-view seats: for each level of risk (the price you pay), the highest return available. Any portfolio below the frontier is the pillar seat — same risk, worse view. You choose where along the frontier to sit based on your own risk appetite, but you never choose a seat below it.

Estimation risk is a recipe with guessed quantities

A recipe for 50 guests needs exact quantities. Suppose the cook has to guess how much of each spice pairs with every other spice — hundreds of guesses. A small error in each adds up, and the biryani goes wrong even though the method was perfect.

MPT is the method. The expected returns, standard deviations and 1,225 correlations for 50 securities are the guessed quantities. The workbook calls the error from these guesses estimation risk: the output of the optimisation can only be as accurate as its inputs.

Key terms in this chapter

Correlation coefficient: Covariance divided by the product of the two standard deviations. It ranges from −1 to +1.
Covariance: A measure of how far two variables move together relative to their own average values.
Covariance terms formula: The number of pairwise covariance terms in a portfolio of n securities: (n² − n) ÷ 2.
Diversification: Reducing portfolio risk by combining assets whose returns are not perfectly correlated, so portfolio risk is below the weighted average risk.
Efficient frontier: The umbrella-shaped curve of portfolios giving the maximum return for each level of risk, or the minimum risk for each level of return.
Estimation risk: The possible error that comes from estimating expected returns, risks and correlations used as inputs to portfolio construction.
Ex-ante return: Expected return found by multiplying each possible return by its probability and adding the results.
Ex-post return: Return calculated from historical data.
Harry Markowitz: Author of "Portfolio Selection" (Journal of Finance, 1952); awarded the Nobel Prize in Economics in 1990 for Portfolio Theory.
Modern Portfolio Theory (MPT): A framework, from Harry Markowitz's 1952 article, for building and choosing portfolios based on expected performance of investments and the investor's risk appetite.
Optimum portfolio: A combination of investments with desirable risk–return characteristics for a given set of constraints.
Perfect negative correlation: Correlation of −1: returns move in opposite directions, giving the complete benefit of diversification and making a zero-risk mix possible.
Perfect positive correlation: Correlation of +1: returns move together in a completely linear way, so combining the assets gives no diversification benefit.
Risk-averse investor: An investor who invests in risk-free options or in options with a positive expected risk premium, and demands more premium for more risk.
Risk aversion index (A): The investor-specific number in the utility formula showing how heavily that investor penalises risk.
Risk premium: The extra return investors demand for bearing risk.
Standard deviation: The square root of variance; the best-known measure of risk, called ex-ante risk when based on expected returns.
Utility score: A number an investor assigns to a portfolio: U = E(r) − ½ × A × σ². Higher return raises it; higher risk lowers it.
Variance: The probability-weighted average of squared differences between possible returns and the expected return. Larger variance means larger risk.
Variance-covariance matrix: The table of variances and covariances needed to calculate risk for portfolios with more than two securities; usually built on a spreadsheet.

Check you have understood it

The L1 test for this chapter has 7 questions. Score 60% to unlock L2 Application (worked numericals and scenarios) and then L3 Exam-Ready (NISM-style traps, timed, with this paper's real marking). Tests, the higher levels and full mock exams need a free account so we can track your progress.

Create a free account

Original explanatory writing based on the official NISM workbook for Series XXI-B, which is published and copyrighted by NISM. Check figures against the current edition at nism.ac.in.

Something look wrong? Report it