Module 4 of 12 · Foundations
Implied versus realised volatility
Implied volatility is a forecast. It is the annualised standard deviation of returns that option prices are assuming for the life of the option. Realised volatility is the standard deviation that then actually occurs. For the major ETFs the exchange publishes a 30-day implied-vol index every day, so the forecast is a public number, and the realised figure can be computed from the fund's own closes 21 trading days later.
On average the forecast runs above the outcome. Across the big equity, gold and oil ETFs, the 30-day index has averaged three to eight points more than the realised volatility that followed, over two decades. That gap is the volatility risk premium. It exists because buyers of options are paying for protection, and the people selling it to them demand to be paid for carrying the tail.
The lab generates a synthetic ETF whose implied vol tracks its true vol plus a premium plus noise, then compares each day's quote with the volatility realised over the following 21 days. The histogram is the point: a mountain to the right of zero and a long tail to the left.
🎛 Vol gap lab
Mean gap
2.6 pts
IV − realised
Median gap
2.7 pts
Months IV was too high
67%
seller's win rate
5th percentile
-11.7 pts
Worst
-52.0 pts
realised far above implied
Synthetic ETF. “Realised” is the volatility that actually happens over the 21 trading days after each quote, annualised. The gap is positive most of the time, which is the premium, and occasionally deeply negative, which is the risk. Changing the premium slider shifts the whole distribution; changing the jump frequency fattens its left tail without moving the typical day much.
The average gap is not the trade; the distribution is
With the premium slider at 3 points, implied vol is too high roughly 75% of the time, and the typical surplus is a few points. The 5th percentile tells a different story: in the bad months realised volatility runs 10 to 20 points above what was implied. A seller who sizes to the average gap and ignores the tail will be right most months and wiped out in one.
Set the jump frequency to zero and the tail disappears; the premium then looks like free money. Real ETFs do jump, and the frequency slider is the honest part of the lab. In the research behind this track, selling the premium on ten ETFs showed a worst month beyond the entire year's income in every decade tested.
Things to try
- • Move the premium slider to 0 and confirm the mean gap goes to zero while the tail stays: the risk is there whether or not you are paid for it.
- • Set jumps to 12 per thousand days and watch the 5th percentile fall even though the median barely moves.
- • Change the seed several times with the same settings: the mean gap is stable, the worst value is not.
Original interactive lesson. Labs run on a seeded synthetic ETF in your browser, built to illustrate mechanics, not to predict any real market. Option prices use a textbook model with zero interest rates. Educational analysis, not investment advice.