Part 1: What volatility is · Chapter 1 of 14 · 9 min
What volatility is and how it is measured
Volatility says how much the price moves, not where it goes. On gold, over 211 market days from December 2025 to October 2026, the daily range (high minus low) has a median of $95.4 and a mean of $117.7, with a minimum of $11 and a maximum of $717; an hour has a median true range of $17. Range, true range and realised volatility from 5-minute data give the same picture (correlation 0.90); the difference between close and open does not (0.55).
How much it moves, not where it goes
Two days can end at the same point and have been lived in opposite ways. In the first the price rises $100 and then comes back $100. In the second it stays still. Close minus open is zero in both, yet someone with a stop $30 from the price was stopped out on the first day and not on the second.
Volatility is this difference: the size of the price's movements over a period, without looking at the direction. It is a measure of distance, in dollars or in percent, and it is the same for an upward move and a downward one. Book 11 used it to size positions, book 12 to build rules. Here we measure it closely, test it and see what can be done with it.
How range and true range are calculated
On a chart candle you read four prices: open, high, low, close. The simplest measure is the range:
> range = high − low
The range does not count the gap between the previous candle's close and this one. If the price opens far above the previous close (a jump), part of the movement stays outside the calculation. That is why the true range (TR) is used: the largest of three distances:
- high − low;
- |high − previous close|;
- |low − previous close|.
An example with simple numbers: a candle has high 108, low 101 and the previous candle closed at 112. The three distances are 7, 4 and 11: the true range is 11, more than the range (7), because the candle began below the previous close.
Read the chart in words
Hourly gold chart with one candle marked with the letter A and, below, the true range bars with the 14-candle ATR line.
| When | What | Price |
|---|---|---|
| 14 Apr 2026 22:00 UTC | Point A | 4,849.97 |
Times in UTC.
On hourly gold the jump between one candle and the next is small: in 97% of the 4,864 candles the true range equals the range. On daily candles, with weekends and news in between, the true range can be much larger.
Four ways to measure volatility
| Measure | How it is calculated | Pros and cons |
|---|---|---|
| Range | high − low of the period | Simple, uses the whole excursion; ignores jumps |
| True range | the largest of the three distances | Counts jumps too; it is the basis of the ATR |
| Standard deviation of returns | typical size of the percentage changes between one close and the next | The most used in finance; ignores what happens inside the period |
| Realised volatility | square root of the sum of squared changes at high frequency (here 5 minutes) | The most complete, but needs dense data |
We calculated them all on the same 211 market days, that is UTC days with at least 18 hourly candles: weekend or holiday days with few trades do not count.
| Measure over the 211 days | Median | Correlation with realised volatility |
|---|---|---|
| Daily range | $95.4 | 0.905 |
| Daily true range | $97.9 | 0.902 |
| Parkinson range (from the ratio of high to low) | $57.5 | 0.907 |
| Absolute value of close minus open | $46.9 | 0.547 |
| Realised volatility from 5 minutes | $58.9 | (reference) |
Three things to read in the table.
- The range measures well. For a random movement the expected range is about 1.6 times the true volatility; here the median ratio between range and realised volatility is 1.58. With four prices per candle you get almost the same result as with hundreds of 5-minute prices.
- Close minus open measures badly. A day that rises and falls by the same amount is worth zero. Whoever judges a day «quiet» because it closed near the open is often wrong.
- You do not need the most sophisticated measure. The correlation between range and realised volatility (0.90) is already high: for trading decisions the range and the true range are enough.
Gold's typical values
For the hour, on the 4,865 hourly candles with the market open (15 December 2025 - 6 October 2026):
| Hourly measure | Value |
|---|---|
| Median true range | $17.0 |
| Mean true range | $22.1 |
| 10th percentile | $8.4 |
| 90th percentile | $39.3 |
| Maximum | $391.5 (29 January 2026, 15:00 UTC) |
| Standard deviation of hourly returns | 0.369% |
For the day, on the 211 days:
| Daily measure | Value |
|---|---|
| Median range | $95.4 |
| Mean range | $117.7 |
| 10th / 90th percentile | $56.7 / $202.2 |
| Minimum | $11.4 (3 April 2026) |
| Maximum | $717 (30 January 2026); then $492 on the 29th and $464 on the 28th |
| Days below half the median range | 10 out of 211 |
| Days above twice the median range | 23 out of 211 |
| Open-to-close standard deviation | 1.90% |
Read the chart in words
Daily gold chart with, below, bars of each day's range and a horizontal line at the median.
The mean ($117.7) sits above the median ($95.4) because the distribution has a long right tail: 23 days are worth more than twice the median, only 10 less than half. We come back to this in chapter 6. For now a practical rule matters: when you report a volatility, write whether it is a mean or a median, because the two differ by a fifth.
Dollars or percent
In dollars the hourly ATR is $18.6 on median, in percent of the price it is 0.41%. In the ten months of data the price lost almost 30% between the January high and the summer lows, so the same volatility in dollars weighs more in percent when the price falls. To compare distant periods look at the percentage; to size a stop in dollars, the value in dollars.
What to do
- Choose a measure and declare it: range or true range, hourly or daily, in dollars or in percent.
- Do not judge a day by the difference between close and open: use the range.
- Write mean or median: the long tail separates them.
- Keep your instrument's typical values (median, 10th and 90th percentile) on one page: they help you recognise a strange day.
- Compare distant periods in percent.
Procedure in four steps
- Choose the measurement period (hour or day) and the price source.
- Calculate the range or true range for each candle.
- Summarise with median, 10th and 90th percentile (and the mean, declared as such).
- Convert it into percent of the price if you compare periods with different prices.
Exercises
Chapter exercises
1. A candle has high 108, low 101 and the previous candle closed at 112. What is the true range?
2. To estimate a day's volatility, what does best in our data?
3. The mean daily range is $117.7 and the median $95.4. What does it mean?
4. Why is it better to look at volatility in percent when comparing January with September?
Have a question or want to share your exercise?
Post in the community, or join the free signals room on Telegram.
Educational content, not financial advice. Trading involves risk.