Markowitz and the Math of Modern Portfolio Theory

This article explores how economist Harry Markowitz revolutionized financial mathematics in 1952 by using mean-variance optimization to create Modern Portfolio Theory (MPT). By translating the abstract concepts of investment reward and risk into the precise statistical measures of mean and variance, Markowitz proved that an asset's risk should not be assessed in isolation. The following sections break down the core mathematical components of his theory—expected return, variance, covariance, and the Efficient Frontier—to explain how Markowitz enabled investors to construct mathematically optimal portfolios.

Quantifying Return and Risk

Before Markowitz published his seminal paper, "Portfolio Selection," investors generally built portfolios by picking individual stocks that offered the highest future returns at the lowest prices. Risk was understood intuitively, not mathematically.

Markowitz changed this by applying statistical formulas to portfolio construction. He posited that investors only care about two things: reward and risk. To put these into an equation, he defined them as follows:

  • Mean (Expected Return): The reward of an asset is calculated as the probability-weighted average of all its possible returns. In a portfolio, the total expected return is simply the weighted average of the expected returns of the individual assets.
  • Variance (Risk): Markowitz defined risk as the volatility of an asset's returns. Statistically, this is measured by variance (or its square root, standard deviation). Variance measures how far an asset's actual returns disperse from its expected mean. High variance equals high unpredictability, which equates to high risk.

The Breakthrough: Covariance

While calculating the expected return of a portfolio is simple addition and multiplication, calculating the variance of a portfolio is much more complex. Markowitz’s greatest insight was proving that a portfolio's total risk is not the weighted average of the risks of its individual assets.

Instead, portfolio risk is heavily dictated by covariance (or correlation). Covariance is a statistical measure of how two assets move in relation to one another.

  • Positive covariance means the assets move in the same direction at the same time.
  • Negative covariance means when one asset rises in value, the other tends to fall.

Markowitz plugged covariance into the mathematical formula for portfolio variance. In doing so, he proved mathematically what the old adage "don't put all your eggs in one basket" meant in practice. By combining assets that have low or negative covariance, the upward volatility of one asset cancels out the downward volatility of another. This allows an investor to drastically reduce the overall variance (risk) of a portfolio without reducing its mean (expected return).

How Mean-Variance Optimization Works

Once Markowitz established the formulas for portfolio mean and portfolio variance, he introduced mean-variance optimization. This is a quantitative process that treats investing as a mathematical optimization problem.

To run a mean-variance optimization, an investor inputs three pieces of data for a chosen group of assets:

  1. The expected return (mean) for each asset.
  2. The variance for each asset.
  3. The covariance Matrix (how every asset moves in relation to every other asset).

Using a mathematical method called quadratic programming, the optimization algorithm tests thousands of different asset weight combinations. The goal of the algorithm is to solve one of two problems:

  1. Maximize the mean for a given level of variance.
  2. Minimize the variance for a given level of mean.

The Efficient Frontier

The output of mean-variance optimization is best understood visually. When you plot the risk (standard deviation) on the horizontal X-axis and the expected return (mean) on the vertical Y-axis, every possible combination of assets forms a dot on the graph.

Markowitz showed that the optimal portfolios—those that offer the highest return for a specific level of risk—form a curved line along the top edge of plotted data. He called this curve the Efficient Frontier.

  • Portfolios on the line are mathematically efficient. You cannot get more return without taking on more risk.
  • Portfolios below the line are inefficient. They expose the investor to unnecessary risk for the return they generate, or they generate too little return for the risk taken.
  • Portfolios above the line are mathematically impossible to achieve given the available assets.

By creating the Efficient Frontier through mean-variance optimization, Markowitz removed guesswork from asset allocation. Investors no longer had to evaluate stocks in a vacuum; instead, they could choose their desired level of risk and use linear algebra and statistics to build the exact portfolio that maximized their returns.