Realised volatility is a measurement
Realised volatility is computed, not quoted. The usual recipe takes the returns between successive closes over a chosen window, measures how widely those returns are spread around their average, and then scales the result so it can be compared across windows. Nothing in that process involves an opinion. Feed in the same closes and you get the same answer every time.
Two choices change the output more than people expect. The window length decides how much of the past is included, so a short window reacts fast and jumps around while a long one is stable and slow. The sampling interval matters too: volatility measured from hourly closes and from daily closes describe different things, because intraday reversals cancel out at the daily level. Any statement about realised volatility is incomplete without both choices attached, and comparing two figures computed on different windows is not a comparison at all.
Implied volatility is a price
Implied volatility comes from the other direction. Someone has paid a premium for an option on gold. A pricing model relates premium to several inputs, most of which are known, and volatility is the one that is not. Solving for the volatility figure that makes the model reproduce the traded premium gives you the implied number. It is extracted from a transaction, so it is a market price wearing the clothes of a statistic.
That has consequences. The figure includes whatever premium sellers demanded for carrying the risk, so it is not a neutral forecast of movement. It also reflects supply and demand for the options themselves. A period where many participants want protection can lift the implied figure without anyone holding a view on how far gold will travel. Reading it as the market expects this much movement is close enough for conversation and wrong in the detail. It is what movement cost, in that window, to the people who traded it.
Looking back and looking forward from the same instant
Place both numbers on a timeline and the distinction becomes obvious. From any given moment, the realised figure describes a window that has ended. The implied figure describes a window that has not started. They are never talking about the same bars, which is why one can be quiet while the other is elevated.
This is why the two can disagree without either being wrong. A market that has drifted for a fortnight will show a low realised figure. If a major scheduled event sits inside the life of the options being quoted, the implied figure can be high at the same time. The low number is a true statement about the past and the high number is a true statement about what the future window costs. Traders who keep both in view treat the first as a description of the current environment and the second as a rough map of where the market expects the environment to change.
The pattern around a scheduled release
The most reliable thing to say about the relationship is structural rather than numerical. Before a known release, the uncertain window sits inside the remaining life of the options being quoted, so the implied figure tends to be elevated. Once the number is out, that uncertainty is resolved and the elevated portion is gone, so the implied figure tends to fall even when price has just moved a long way.
Realised volatility does the opposite. It cannot rise before the event, because nothing has happened yet. It rises on and after the release as the large bars enter the measurement window, and then it decays as those bars age out. The two curves therefore cross around the event in a characteristic way, which catches out traders who expect a big move to leave volatility high everywhere. One measure is backward looking and one has just lost its reason to be elevated. The surrounding routine is covered in the piece on gold around rate decisions.
Why the gap between them is watched
The difference between what a window cost in advance and what it actually delivered is the one comparison that uses both numbers properly. If movement priced richly and then did not arrive, sellers of that risk were paid for something that did not happen. If it priced cheaply and the market tore apart, buyers were underpaying. Describing that gap after the fact is straightforward bookkeeping.
Where it stops being straightforward is prediction. Knowing that the gap has had a tendency in one direction historically is not the same as knowing it will on the next event, and the circumstances that produce an elevated implied figure are not identical from one release to the next. Options traders build whole positions around this comparison, with position sizing and hedging rules to survive being wrong. A chart trader borrowing the idea should borrow only the framing: expensive expectation and quiet outcome is one state of the world, and cheap expectation and violent outcome is another.
What a chart only trader can use instead
Most traders reading a gold chart have no option data in front of them, and that is workable as long as the limitation is stated. Everything available from price alone is a realised measure. Average true range describes the size of recent bar ranges. The spread of the last several sessions describes how much ground is being covered. A standardised stretch reading says how unusual the current range is against its own recent history, which is the idea behind range stretch measures on gold.
These are honest and they are all backward looking. None of them contains forward information, so they cannot warn that a quiet market is about to become violent. The one piece of forward information a chart trader genuinely has is the calendar, because the dates of scheduled releases are known in advance. Marking those dates is the practical substitute for watching an implied figure, and it is most of the benefit at none of the cost.
What changes when the regime changes
Volatility regime should change mechanics, not conviction. When bar ranges expand, a stop placed at a quiet market distance becomes a stop sitting inside ordinary noise, and the only ways to respond are a wider invalidation with smaller size, or standing aside. Keeping the same distance and the same size is the choice that quietly increases risk. The arithmetic is laid out in risk management on gold.
The other mechanical change is execution cost. Quotes widen when movement arrives, so the same idea costs more to enter and exit, and the extra cost lands at the worst moment. Marking the current range regime before the session, rather than reacting to it mid trade, is what the Market Regime label on the live chart is for. The transaction side of this is covered in spread and slippage around gold news.
Where both readings mislead
Realised volatility misleads at turning points, by design. It is an average over a window, so it is lowest immediately before an environment changes and highest immediately after the change has already been paid for. Anyone sizing a position from a realised figure in a long quiet stretch is sizing from the most reassuring number the market will produce, right at the point it is least informative.
Implied volatility misleads in its own way. It carries a risk premium, so it tends to sit above what eventually shows up, which makes it look like a market that constantly overestimates danger. It also reacts to demand for options rather than to a view on gold, so a figure can move on flows alone. And a single implied number flattens a whole structure of different expiries and strikes into one quote. The practical conclusion is unglamorous: use realised to describe the environment you are in, use the calendar to know where it may change, and resist turning either number into a forecast.
FAQ
Is implied volatility a forecast of how much gold will move?
Not exactly. It is the volatility figure that makes a pricing model match a traded option premium, so it reflects what buyers and sellers agreed on, including the premium demanded for carrying risk. It is better described as the cost of movement in that window than as a neutral prediction.
Why does implied volatility fall after a big release even when gold moved a lot?
Because the uncertainty it was pricing has been resolved. Before the release, the unknown outcome sat inside the option life and lifted the premium. Once the number is public, that component disappears. The realised figure rises at the same time, since the large bars have now entered the measurement window.
Does high realised volatility mean gold is trending?
No. Realised volatility measures how widely returns are spread, not whether they point the same way. A market can swing violently in both directions and produce a high figure with no net progress. Deciding whether movement is one sided needs a separate measure such as a directional strength reading.
Should I widen my stop when volatility rises?
The mechanical point is that a fixed distance means something different in each regime. If ordinary bar ranges have expanded, a distance that worked in a quiet market now sits inside normal noise. The usual responses are a wider invalidation with reduced size, or not taking the trade, rather than keeping both unchanged.
Can I work out implied volatility from a price chart?
No. It can only be extracted from traded option premiums, so a chart alone cannot produce it. Everything computed from price history, including average true range and range based stretch readings, is a realised measure and contains no forward looking component at all.
ⓘ See these ideas on real price: open the free XAUUSD live chart.