Charts See the terminal

Volatility O-H-L-C

An annualised volatility estimate that uses each bar's open, high, low and close, so it reads more from each bar than a close-only measure.

Volatility Add to chart

Live Volatility O-H-L-C preview · BTC/USDT · Binance

What is Volatility O-H-L-C?

Volatility O-H-L-C estimates annualised volatility using the Garman-Klass method. Instead of looking only at closes, it uses all four prices of each bar, the range from high to low and the move from open to close, so it picks up how far price travelled inside the bar. The result is averaged over Length bars, scaled to a year and plotted in its own pane.

It is an estimate of how much price has been varying, not a forecast and not a direction. Its assumptions about how price moves inside a bar do not always hold, and when they fail the reading is off rather than merely noisy.

How to use Volatility O-H-L-C

  • Rising line. Recent bars have had wider ranges relative to price. Volatility is increasing, whichever way price is moving.
  • Falling or low line. Bars are getting tighter. The market is quietening.
  • Reading the number. The value is an annualised percentage. A reading of 60 means about 60% annualised volatility.
  • Against close-to-close. Plotted next to Volatility Close-to-Close, this line is usually steadier for the same window, because each bar contributes more information.

These are ways to read volatility, not entry signals on their own.

Settings

SettingDefaultWhat it changes
Length10 Bars the per-bar estimate is averaged over. Shorter reacts sooner; longer is smoother. Minimum 1.
Market Closed Percentage0 On charts below 1D, the share of each day the market is closed, which lowers the number of bars counted per day when annualising. 0 fits a market that trades around the clock. No effect on 1D and higher. From 0 to 100.
Days Per Year252 The number of days treated as one year when annualising. 252 is a trading-day count; crypto trades every day, so 365 is the natural figure for a crypto market. It only rescales the line. From 1 to 100000.

Adjust the averaging window, the closed share of the day and the annualisation factor in the indicator settings.

Calculation

For each bar, the Garman-Klass term combines the log of the high-low range with the log of the open-close move. The indicator averages that and scales it to a year:

range = ln(high / low)
body  = ln(close / open)
gk    = 0.5 * range^2 - (2 * ln(2) - 1) * body^2

bars per day = 86400 / bar length in seconds * (1 - Market Closed % / 100)   below 1D
             = 1                                                         on 1D and higher
vol   = 100 * sqrt( SMA(gk, 10) * 252 * bars per day )

The constant 2 * ln(2) - 1, about 0.386, is not a tuning value. It comes from the method's model of price as a continuous random walk with no drift and no gaps, and the method depends on that model holding. A 24-hour market such as crypto has fewer gaps than an exchange that closes, which suits it better.

The second term is subtracted, so a bar that closes far from its open relative to its range can give a negative term. The indicator does not guard against this. If the window's average turns negative, the square root has no value and the line shows a gap for that bar.

The result is a percentage. On charts below 1D the per-bar figure is scaled by the bars in a day, 24 on 1h, and then by Days Per Year, so the level is comparable across intraday timeframes. On 1D and higher each bar counts as one day, which holds on a daily chart; on weekly or longer bars the level is not a true annual figure.

Limitations

  • It rests on a model. The estimator assumes price wanders randomly without drift or gaps. Strong trends, halts, gaps and thin order books break that assumption.
  • Gaps in the line. When bars close far from their open relative to their range, the window average can turn negative and the line shows no value.
  • The annual figure depends on timeframe. Intraday bars are scaled by the bars in a day and daily bars count as one day each, but weekly and longer bars also count as one day, so the level on those charts is not a true annual figure.
  • The live candle moves. The current bar's high, low and close can still change, so its value moves until that bar closes.
  • It is not a strategy. It measures volatility. It gives no entry, exit, sizing or risk management on its own.

Frequently asked questions

It is an annualised volatility estimate using the Garman-Klass method, which reads the open, high, low and close of every bar. Because the high and low show how far price travelled within the bar, it extracts more information per bar than a measure based on closes alone.

The line is an annualised percentage, so 40 means about 40% annualised volatility. On intraday charts the per-bar value is scaled by the bars in a day, 24 on 1h, and then by Days Per Year; on 1D each bar counts as one day. Read rises and falls against the line's own history on the same chart and timeframe.

Yes. The indicator declares three inputs: Length, default 10 with a minimum of 1; Market Closed Percentage, default 0, from 0 to 100; and Days Per Year, default 252, from 1 to 100000. Open the indicator's settings on the chart to change them. For a crypto market, which trades every day, 365 days per year matches the calendar.

No. Each value uses only that bar and the ones before it, so values on closed bars are not revised. The current candle's reading moves while that candle is live and settles when it closes.

The per-bar term subtracts the size of the open-to-close move. If bars in the window close far from their open compared with their range, the average can go below zero, and the square root of a negative number has no value. The indicator leaves that bar blank rather than inventing a figure.

Add to chart