What volatility actually measures

Volatility is a measure of how far a price tends to move, in either direction. It says nothing about whether the next move is up or down. A market that rises 5% one day and falls 5% the next is volatile, even if it ends the week where it started.

The standard version is realized volatility, sometimes called historical volatility. You take a run of daily price changes, work out their standard deviation, and scale it to a year so different assets and different windows can be compared on one yardstick. When someone says Bitcoin's volatility is 40%, they almost always mean this annualized number.

There is a second kind. Implied volatility is backed out of option prices and tells you how much movement options traders are paying to protect against. The VIX does this for the S&P 500, and Deribit publishes a similar index for Bitcoin called DVOL. Realized volatility looks back. Implied volatility is a market price for the future. This guide is about the first one, because anyone can calculate it from free data.

How to calculate it, step by step

Here is the method we use throughout this article, applied to Binance's BTCUSDT daily closes:

  1. For each day, take the log return: the natural log of today's close divided by yesterday's close. For small moves this is almost the same as the percentage change.
  2. Pick a window, usually the last 30 days, and compute the standard deviation of those returns.
  3. Multiply by the square root of the number of trading days in a year. Crypto trades every day, so that is 365, and the square root is about 19.1. Stock markets use 252 trading days, with a square root of about 15.9.
Annualized volatility = stdev(ln(Close_t / Close_t-1)) × √365

To turn the annual number back into something you can feel, divide instead of multiplying. On September 27, 2026, Bitcoin's 30-day realized volatility on our count was 40.4%. Divided by 19.1, that is a typical daily move of about 2.1%. With Bitcoin closing that day at $84,472, one ordinary day's swing was worth roughly $1,790. Over a 30-day month the same volatility implies a typical move of about 11.6%.

"Typical" here means one standard deviation. Roughly two days in three should stay inside that band if returns were well behaved. Bitcoin's are not. Its big days come more often than a bell curve predicts, which is the practical reason volatility numbers should be read as a floor on risk rather than a ceiling.

Nine years of Bitcoin volatility, measured

We downloaded every daily candle for BTCUSDT from Binance's public API, from its first day of trading on August 17, 2017, to the close on September 27, 2026: 3,328 daily returns. Across the whole period, annualized volatility was 67.5%. Year by year it looks like this:

YearAnnualized volatilityDays moving 5% or moreDays moving 10% or more
201883.9%7719
201968.1%4511
202080.8%428
202180.8%7410
202264.3%409
202343.5%211
202452.6%262
202541.6%130
2026, to Sep 2746.2%14 in 270 days2

The trend is hard to miss. Bitcoin's volatility has roughly halved since 2018, and the number of days with a 5% move fell from 77 to 13. 2025 was the first full year in the data without a single 10% day, and it included the record high: the Binance spot pair reached $126,199.63 on October 6, 2025.

The extremes are worth knowing too. The highest 30-day reading was 208.2% on April 6, 2020, after the pandemic crash of March 12, 2020, when Bitcoin fell 39.5% in a single day. The lowest was 16.5% on August 13, 2023, a stretch so quiet that it sat below the long-term median of the stock market's VIX, which we come to below.

2026 has not been smooth. On February 5 Bitcoin fell 14.0% to close at $62,910, and on February 6 it bounced 12.2%. The 30-day reading peaked at 82.5% on March 4. It then drifted down to 21.6% on August 15, the calmest point of the year, just before three strong up days from August 19 to 21.

Bitcoin against Ethereum, gold and stocks

A volatility number only means something next to another one. We ran the same calculation on Ethereum and on PAX Gold (PAXG), a token backed by one troy ounce of physical gold per token and held by Paxos. PAXG trades on Binance around the clock, which lets us measure gold with exactly the same method and the same 365-day convention. It is a proxy, not the spot gold price, and it can drift slightly from it.

YearBitcoinEthereumGold (PAXG)Bitcoin ÷ gold
202180.8%108.5%15.2%5.3x
202264.3%87.3%15.9%4.0x
202343.5%46.5%12.5%3.5x
202452.6%64.0%15.5%3.4x
202541.6%74.3%20.2%2.1x
2026, to Sep 2746.2%62.4%31.4%1.5x

Ethereum has been more volatile than Bitcoin in every year, and in 2025 the gap was the widest in the data, 74.3% against 41.6%.

The bigger surprise is gold. In 2021 Bitcoin swung more than five times as hard as the gold token. So far in 2026 the ratio is 1.5 times. Most of that change came from gold's side. Gold's 2026 reading of 31.4% is the highest for any year since PAXG started trading on Binance in 2020, and in January 2026 the gold token was more volatile than Bitcoin: 48.2% against 42.6%. That month ended with a crash. Kitco reported that gold fell more than 10% on Friday, January 30, the biggest one-day selloff in its history, two days after its biggest one-day gain. On Binance, PAXG rose 6.7% on January 28 and fell 10.1% on January 30.

For stocks we do not have a like-for-like series in this piece, so we lean on a reported figure. Cboe gives the VIX a long-term median of 17.6. The VIX is implied rather than realized volatility, so the comparison is rough. Still, Bitcoin's median 30-day realized volatility across nine years is 54.3%, about three times that level. Bitcoin is calmer than it was, but by any of these yardsticks it is still far from a calm asset.

Where volatility sits now

At 40.4% on September 27, Bitcoin's 30-day volatility is lower than about three quarters of all 3,299 daily readings since 2017. It sits at the 23rd percentile. The 90-day figure is similar, at 38.7%.

A low reading does not mean nothing is happening. Six of 2026's fourteen days with a 5% move came between August 19 and September 21, and all six were up days. A market can post several large moves in a few weeks and still show a modest 30-day number, because the calm days around them pull the average down. That is one reason to check more than one window before drawing a conclusion.

Does a quiet market come before a big move?

You will often read that low volatility is a coiled spring. We tested the idea on the same Binance data. For every day with a full 30 days of history behind it and 30 days ahead, we recorded the 30-day volatility and then the size of Bitcoin's move over the next 30 days.

DaysMedian size of next 30-day moveNext move 10% or moreHigher after 30 days
All 3,269 days11.7%56%55%
Quietest 10%, volatility 31.7% or less15.2%64%50%
Wildest 10%, volatility 92.9% or more17.1%69%54%

After the quietest days, the next month's move was larger than usual: a median of 15.2% against 11.7% for all days. So the folklore has something to it in this sample. But look at the last column. After quiet periods, Bitcoin was higher 30 days later exactly half the time. Quiet told you a move was somewhat more likely. It told you nothing about which way.

Two cautions. The windows overlap, so neighbouring days are not independent tests. And the quiet days cluster heavily in recent years: 253 of the 327 fall in 2023, 2025 and 2026. The honest reading is a handful of episodes, not hundreds of trials.

The steadier pattern is clustering. The correlation between today's 30-day volatility and the volatility of the following 30 days is 0.46. Turbulent months tend to follow turbulent months, and calm months tend to follow calm ones. The link is loose enough that it works as a description of the current regime, not as a timing tool.

How to use volatility in practice

Size positions to the current number, not last year's. If Bitcoin's typical day is 2.1%, a stop placed 1% away will be hit by ordinary noise most weeks. Our stop loss guide shows how to work out position size from the distance to your stop. When volatility doubles, the same risk in dollars means half the position.

Remember what leverage does to it. A 2% daily move becomes 20% of your margin at 10x. That is how ordinary volatility turns into liquidations, and why the cost of holding leverage, covered in our funding rates guide, is only part of the risk.

Compare like with like. Check whether a quoted figure uses 365 or 252 days, a 30-day or a one-year window, and realized or implied volatility. The same week of Bitcoin trading can produce very different headline numbers depending on those choices.

Treat low volatility as a warning about size, not a signal about direction. In our test, quiet months were followed by larger moves, and they were up only half the time.

Volatility also matters for longer-term investors. A strategy like dollar-cost averaging exists precisely because an asset this volatile makes the timing of a single purchase a large part of the outcome.