Realized volatility
Realized volatility is the standard deviation of an asset's past returns over a set window, usually annualized. It measures how much price actually moved, unlike implied volatility, which is the movement priced into options.
Bitcoin volatility now
16:44 UTCHow it works
Realized volatility, also called historical volatility, is calculated directly from price history:
- Take closing prices over a window, for example the last 7 or 30 days.
- Compute each period's log return: ln(today's close / yesterday's close).
- Take the standard deviation of those returns. That is the daily realized volatility.
- Annualize it by multiplying by the square root of the number of periods per year. Crypto trades every day, so the factor is √365 ≈ 19.1, not the √252 used for stocks.
The annualized figure is a convention for comparing assets and windows. It is easier to think in daily moves:
| Daily standard deviation | Annualized |
|---|---|
| 1% | about 19% |
| 2% | about 38% |
| 3% | about 57% |
| 4% | about 76% |
To go the other way, divide by 19.1. An annualized realized volatility of 50% corresponds to a typical daily move of about 2.6%.
The window matters. A 7-day figure reacts within days to a burst of movement; a 30-day figure smooths it out and takes weeks to come back down.
A worked example
Eight daily BTC closes: $100,000, $101,500, $99,800, $100,900, $98,700, $99,600, $102,300, $101,200.
The seven daily log returns are +1.49%, −1.69%, +1.10%, −2.20%, +0.91%, +2.68% and −1.08%.
- Standard deviation of those returns (sample): 1.83% per day
- Annualized: 1.83% × 19.1 = about 35%
Price ended the week only 1.2% above where it started, yet realized volatility was 35%. Volatility measures the size of the moves, not the net change.
What it tells you
The box above shows Bitcoin's annualized realized volatility over 7 and 30 days, along with the typical daily move and the range over the last 24 hours. When the 7-day figure is well below the 30-day, the market has gone quieter recently. When it is well above, a burst of movement has just happened and the longer window has not caught up yet.
Realized volatility is a description of the past. It is the right input for asking how large recent moves were, and a weak input for how large the next ones will be. That forward-looking question is what implied volatility answers: the volatility that option prices imply, published for BTC and ETH as Deribit's DVOL index. Comparing the two is more useful than either alone:
- Implied above realized is the normal state. Option sellers charge a premium for carrying the risk of a jump, and on average implied volatility has run above realized.
- Implied rising while realized stays flat means the options market is pricing in movement that price has not shown yet, often ahead of a scheduled event.
- Realized jumping above implied means moves are larger than options priced in. Option sellers lose, and implied volatility usually reprices upward quickly.
Low realized volatility also changes behavior. After weeks of small moves, stops get tighter and leverage creeps up because the recent range feels safe. That builds the clustered positioning behind a liquidation cascade.
When it matters most
Realized volatility matters most at regime changes. Because it is measured after the fact, it registers a breakout only once the move is underway, and it describes the old regime right up to the moment it stops being true. Position sizes and stop distances based on a calm 30-day window are sized for a market that may already be gone.
It is also the right tool for sizing. A position that risks 2% on a stop placed one typical daily move away needs a smaller size when that typical move is 4% than when it is 1.5%. Scaling position size to realized volatility keeps the risk per trade roughly constant as the market regime changes.
Common misconceptions
"Low volatility means low risk." It means recent moves were small. Long calm stretches often coincide with rising leverage and cheap options, which makes the eventual move larger.
"Volatility measures how much price went up or down." It measures the size of the moves around the average, regardless of direction. A week that ends flat can have high realized volatility.
"Unrealized volatility is a separate measure." It is informal shorthand for movement that has not happened yet. The established pair is realized versus implied volatility.
"A 1-sigma daily move covers almost every day." Under a normal distribution it covers about two days in three. Crypto returns have fat tails, so days of three or more standard deviations happen far more often than a normal distribution predicts.