Why Options Usually Cost More Than the Moves They Predict

Two forecasts, one market

Imagine two weather forecasters. One watches the sky and calculates what actually happened. The other reads what people are willing to pay for umbrellas. Both are talking about the same storm, but they are answering different questions. That is roughly the situation in options markets, and it explains one of the most durable patterns in finance.

Start with the vocabulary, because these terms get tangled fast. Volatility is simply a measure of how much an asset's price moves around over time. Statisticians split it into two families. Realised volatility — sometimes called historical volatility — is calculated from actual observed returns over a set window, typically the standard deviation of those returns. Implied volatility runs the other way: it is the volatility number you have to plug into an option pricing model to make the model's price match the price actually quoted in the market. One looks backward. The other is baked into what people are paying today for the right to buy or sell something later.

When you compare the two, a pattern shows up again and again: implied volatility tends to sit above the realised volatility that eventually unfolds. That gap is the volatility risk premium. Think of it as the fee people pay to hand off uncertainty to someone else.

Why the gap exists at all

If markets were frictionless and everyone were indifferent to risk, the two numbers should roughly converge over time. They don't, and the reason is human rather than mechanical.

Options are, in effect, insurance contracts. A put option protects a portfolio against a fall. Many investors dislike losses far more than they enjoy equivalent gains, so they are willing to pay up for that protection. Sellers of options — the ones providing the insurance — demand compensation for taking on the risk. That compensation shows up as elevated implied volatility. Researchers have described the implied-versus-expected variance difference as a proxy for risk aversion, and others tie it to broader macroeconomic uncertainty that markets price but cannot easily observe.

There is a second, subtler mechanism. Volatility tends to rise when prices fall, and fall when prices rise. That asymmetry gives option sellers a genuine hedging service to provide: they absorb losses precisely when buyers need the payout most. Risk that hurts in bad times commands a higher price than risk that only bites in good times. A fair premium is the natural result.

The premium is not a constant. It moves. It widens when fear spikes and narrows in calm stretches. And notably, it is not confined to equities. Studies of variance risk premia in government bond and credit markets find a common short-run component that predicts returns across equity, bond and credit markets — a shared heartbeat of uncertainty.

The term structure: not one premium, but many

Here is a detail that surprises people. The premium is not the same for every expiry date.

Research estimating how investors price variance risk across different option maturities finds that the premium declines in absolute size as maturity lengthens, but remains meaningfully different from zero out to around nine months. In plain terms: the compensation for bearing volatility risk is richest at short horizons and thins out as you look further ahead. During stressed periods the required compensation rises and the curve steepens.

That shape tells you something practical. When someone quotes you "the" volatility risk premium, they are quietly picking a maturity. A thirty-day reading and a nine-month reading can tell different stories about the same market.

FX and crypto: same idea, louder numbers

In foreign exchange, volatility behaves with its own personality. Currencies are relative prices — every exchange rate involves two economies, two central banks and two policy paths. When uncertainty about rates or trade policy rises, the FX options market reprices that uncertainty quickly. Forward-looking gauges of currency volatility move together with options demand, and in higher-volatility environments hedging activity becomes a dominant feature of the market rather than a background hum.

Crypto options sit at the far end of the spectrum. Bitcoin is dramatically more volatile than the S&P 500, and its variance risk premium is correspondingly larger. Research on Bitcoin options traded on major derivatives venues finds the premium varies with the market's volatility regime: in calm periods, a relatively large share of Bitcoin's return premium comes from the upside, and the variance risk premium is high. In turbulent periods, upside and downside contributions balance out more evenly and the variance risk premium compresses. That is a genuinely different pattern from what equity markets show.

Bitcoin also throws a curveball at the textbook. The Black–Scholes model assumes volatility is constant and returns are log-normal — tidy assumptions that real markets cheerfully violate. The result is the volatility smile: implied volatilities differ across strike prices instead of forming one flat number. Bitcoin options display a forward skew that has led some researchers to classify Bitcoin closer to a commodity than a currency for option pricing purposes.

One more complication worth naming honestly: estimating implied volatility in these markets is not plug-and-chug. There is no neat formula that reverses the option price into volatility, so practitioners use numerical root-finding methods — iterative techniques like Newton-Raphson or bisection — that converge on an answer. Different methods converge at different speeds and can land on slightly different numbers. The measurement itself carries uncertainty.

How to actually measure it

You need two ingredients, and they must be compared over the same horizon or the exercise is meaningless.

First, the implied side. For major equity indices, the best-known gauge is the VIX, a model-free measure built from a broad array of call and put premiums with thirty days to expiry across many strike prices. FX and crypto markets have their own equivalents, though liquidity varies enormously by currency pair and venue.

Second, the realised side. The most accurate approach uses high-frequency intraday returns, because frequent sampling captures the true variance far better than daily closing prices. Market microstructure frictions — bid-ask bounce, thin trading, stale quotes — limit how finely you can sample before the noise outweighs the signal.

Then subtract. The difference between the ex-ante risk-neutral expectation of variance and the statistical expectation of realised variance over the same window is your estimate of the premium. Everyone agrees on the concept. Nobody agrees on the best way to estimate the expected leg. Some use multi-lag autoregressive models, others use lagged implied and realised variance, and the choice changes your answer. A premium estimated with one method may look modest; with another, substantial.

The honest summary: the volatility risk premium is real, well documented across asset classes, and reliably hard to pin down to two decimal places. Treat any single number with appropriate scepticism, and remember that regulation, market structure and available instruments differ by jurisdiction — what is straightforward in one market may be restricted or impractical in another.

What the premium is not

It is tempting to read all this and conclude that selling options is free money. It is not, and the research is unusually blunt about it.

The return distribution from selling volatility is severely abnormal. Historical analysis of short volatility strategies shows put sellers could incur losses reaching several hundred percent, and there is strong serial correlation in large negative days from the seller's perspective. Substantial margin reserves are required, which drags reported returns down considerably. Selling volatility is closer to selling catastrophe insurance than to clipping coupons: many quiet premium collections, punctuated by occasional very bad days.

That does not make the premium unreal. It makes it compensation for genuine risk. Understanding the difference between those two statements is most of the education.