Why Volatility Is Not the Same as Risk

Portfolio optimization frameworks treat volatility and risk as equivalent. Risk reports measure them in the same units. And yet the two concepts may describe different things. Volatility measures how much prices vary over a given period. Risk, at least in the sense that matters to investors with multi-year investment horizons, has more to do with the probability of permanent loss in portfolio value. The distinction may seem trivial or semantic, the data suggests it has practical consequences for how portfolios are constructed and how losses are evaluated.

Consider the CBOE Volatility Index. The VIX is derived from S&P 500 index options. Its calculation aggregates prices across a wide strip of out-of-the-money puts and calls with approximately 30 days to expiration. Each option contributes to the estimate in proportion to its distance from the current index level, weighted by the inverse of the squared strike price. The resulting figure is the market's implied estimate of annualized 30-day variance, expressed as a percentage after taking the square root (CBOE, 2019).

The description reveals the true meaning underlying VIX. It is not a forecast. It is a price. It reflects what option market participants are collectively willing to pay for convexity over the next month. And like any price, it is shaped by supply and demand as much as by information. When portfolio insurers buy puts in size, the VIX rises. When structured product dealers sell volatility, it falls. The signal is real, but may not be purely informational.

If volatility were a reliable forward-looking measure of risk, then periods of high implied volatility should be followed by poor returns. However, between 1990 and 2024, months in which the VIX closed above 30 were followed by average 12-month S&P 500 returns of approximately 21%. Months in which the VIX closed below 15 were followed by average 12-month returns of approximately 8%.

This pattern makes more sense if we separate the two ideas. High option-implied volatility tends to coincide with depressed prices, and depressed prices tend to produce higher subsequent returns. The VIX is reflecting the temperature of the market, not the fundamental riskiness of the underlying businesses. An investor who sold equities every time the VIX spiked above 30, treating it as a risk signal, would have systematically exited at some of the most attractive entry points of the past three decades.

In several contexts, volatility displays characteristics that diverge from common understanding of investment risk. Adding a risk-free asset to a portfolio, for example, does not change the measured volatility of the equity component. But it improves the beta of the overall position. Volatility measures dispersion around an average return, not the investor’s preference between upside and downside outcomes. It is, therefore, symmetric by construction In log-return terms, a doubling and a halving are equal-magnitude moves in opposite directions, and standard deviation treats them accordingly.

The risk aggregation properties are also worth thinking about. Two uncorrelated assets, when combined, may show lower portfolio volatility than either asset alone. This is the basis of diversification. But the reduction depends on correlation remaining stable, and Longin and Solnik (2001) showed that equity correlations tend to increase during downturns. The measured diversification benefit can shrink precisely when it is most needed. A portfolio that appeared well-diversified on a volatility basis in 2007 may have behaved very differently in 2008.

Then there is the question of tails. A portfolio with small daily fluctuations and occasional 30% drawdowns has low measured volatility most of the time. The standard deviation estimate would not have warned you about the drawdown in advance, and it would have recovered quickly once the drawdown ended. The risk was there. The measure missed it.

None of this makes volatility useless. It remains a practical input for position sizing, option pricing, and relative value comparisons. But it captures the dispersion of recent returns, not the probability of outcomes that haven't happened yet. Risk, as it affects an investor's ability to compound capital over a decade or more, lives in leverage, in concentration, in business model fragility, in liquidity mismatches, and in assumptions that have not yet been stressed. These things do not map reliably to price fluctuations. A stock can be calm and fragile. Another can be volatile and fundamentally sound. Knowing which is which requires judgment that no single statistic can replace.