2. Implied volatility is a price, not a forecast

Derivatives 3: Volatility

This is the level's central module, and it can be summed up in a sentence almost everybody repeats wrongly: implied volatility does not predict what is going to happen, it prices what it costs to protect yourself against it happening. Those are two very different claims and they lead to opposite decisions.

What the buyer of an option is really buying

He is buying a transfer of risk. Just as a fire insurance policy hands the insurer the ugly part of the distribution of outcomes for your house, an option hands the seller the ugly part of the distribution of outcomes for your position.

The comparison is not a teaching device: it is the exact structure of the contract, and the important thing follows from it. The premium an insurer charges is not its forecast about whether your house will burn down. It is the expected loss plus a margin that compensates for the capital tied up, the uncertainty and the fact that claims arrive in clusters. The forecast is one ingredient of the price, and not the one that calls the shots when things get tense.

The implied has been systematically above what actually happened

This is the empirical observation on which a good part of the options business rests. You note down an index's thirty-day implied, wait a month, work out the volatility that was actually realized and compare the two numbers. Then you repeat the exercise thousands of times.

The result, on the S&P 500 and over the last three decades, is that the implied has exceeded the subsequent realized in the great majority of periods, with an average difference of the order of three or four points. This is not a flaw nobody has spotted: it is a payment, and it has a name, the volatility risk premium. It is what the buyer of protection accepts paying over the odds in exchange for sleeping soundly.

That positive average is not easy money, because it hides a very lopsided distribution: many months of modest income and the occasional month in which the realized far exceeds the implied. What happens to somebody who takes the sell side is the subject of a later level; here it is enough to recognize the phenomenon.

"The implied is high" does not mean "there is going to be a move"

Out of the previous section comes the most useful reading correction of the level. When a headline says implied volatility has shot up, it is saying that protection has become more expensive. That may be because a move is expected, yes. It may also be because there is more demand for cover than supply, because whoever would sell it is constrained by their margin, or because there is an event with a date in the calendar.

And there is something more important. Even if the implied got it right and there were a move, that is not enough to make money. The buyer does not profit from the move: he profits from the difference between the one he paid for and the one he received. Buying with an implied of 40% and finding a realized of 30% is losing money, even though 30% annualized is an enormous amount. The bar is not zero: it is the price you paid.

Getting the direction right and losing money anyway

The most painful case, and the one that most surprises people arriving from the cash market. It happens around events with known dates: quarterly results, a court decision, a macroeconomic release, a regulatory approval.

In the preceding days, the implied of the expiry containing the event swells, because there is a great deal of uncertainty concentrated on one date. When the event takes place, that uncertainty disappears all at once. The implied deflates in minutes, and the option's price takes two hits of opposite sign: the one from the underlying's move and the one from the collapse in the implied. An option's sensitivity to the implied has a name of its own, vega, and it is the subject of the next level.

An example with round numbers, worked out with the classic model and purely illustrative. A share trades at 100 on the eve of its results. A call option with a strike of 105 and three weeks of life has an implied of 70% and is worth around 4.60. The accounts are published, they are good, the share rises to 103 and the implied sinks to 35%, its normal level with no events ahead. That option goes on to be worth around 2.50.

The investor got the direction right, the share rose 3% in a session, and he lost close to half. There is no arithmetic error: he paid for uncertainty and the uncertainty was resolved.

How to read the implied without falling into the trap

• Compare each asset with itself. An implied of 45 can be cheap in a biotech and extremely dear in a utility. What is informative is where it sits relative to its own range over the past year.
• Separate the level from the event. An implied that is high because of a jumpy regime does not behave like one that is high because of a date in the calendar, which will deflate on its own.
• Work out the move you need. Translate the implied into an expected move and ask yourself whether the one you expect is bigger. If it is not, you get the direction right and lose.
• Dear is not the same as bad. Expensive protection can be the right decision if it covers something you cannot afford. It is worth knowing that you are overpaying, and why.

A real case: November 2016

In the days before the American presidential election of November 8, 2016, the S&P 500's thirty-day implied climbed above 22 points, well above its level in those weeks. The result was the one almost nobody was pricing in. And yet, two sessions after it became known, that same implied had fallen below 15, with the market rallying hard.

Anybody who had bought protection was completely right: the result was a surprise. He lost money anyway. He had paid a price made of uncertainty about the outcome, and once there was an outcome that price ceased to exist. The implied never promised that the market was going to fall. It only said what the insurance cost on the eve.