Volatility Toolkit

// Always use British English

The Information Hidden in Range

🧪 Open Source Tool

Most volatility measures look only at closing prices and quietly discard the rest of what each bar already revealed — its high, its low, its open. The Volatility Toolkit is built around a better idea: range-based estimators that read the whole candle. By using all four OHLC prices, these methods extract several times more information from the same data than a close-to-close calculation, producing stable estimates from far fewer bars.

Tip

Open Source Philosophy: Both the Volatility Toolkit indicator and the underlying VolatilityToolkit Library are fully open source. They build on the FiltersToolkit Library for smoothing and the ColourUtilities library for palettes — the same components we use across our own systems, shared so you can study, verify, and extend them.

Filters Toolkit The companion DSP filter laboratory that supplies this toolkit's smoothing options.

Ultimate Smoother The zero-lag filter behind the smoothing that avoids volatility drop-off artefacts.

Kinetic Trend Envelope The trailing-stop component that consumes five of these estimators to size its bands.

Filter Methods Mathematical foundations of the DSP filters used throughout the suite.

Why Range Beats the Close

A classical close-to-close estimate treats every bar as a single number and throws away the path the price actually travelled to get there. Academic research since 1980 has repeatedly shown that the intrabar range carries most of the information about the underlying diffusion — so estimators that read the high, low, and open are dramatically more efficient.

Statistical efficiency here has a precise meaning:

• Parkinson (1980) is roughly 5× as efficient as close-to-close — the same accuracy from about 80% less data.
• Garman-Klass (1980) reaches about 7.4× by adding the open and close.
• Yang-Zhang (2000) achieves up to 14× while also handling drift and overnight gaps.

Efficient does not mean narrower

Higher efficiency describes lower sampling variance under the estimator’s assumptions — not a guarantee of a tighter reading. A gap-aware estimator can legitimately widen where a pure high-low estimator stays calm, and a drift-independent estimator behaves differently during sustained trends. Match the estimator to the market’s structure, then validate on your own instrument.

Three Scales, One Rule

Volatility can be expressed in different units, and mixing them silently produces nonsense. The toolkit groups every estimator by its output scale and enforces the rule that only same-scale estimators may be compared or aggregated — warning you whenever a configuration would mix them.

Scale group	Unit	Estimators	Comparable & aggregatable?
Log-return (σ)	Standard deviation of log returns	Close-to-Close, Parkinson, Garman-Klass, Rogers-Satchell, Yang-Zhang	Yes — and annualisable via the √time rule
Price ($)	Absolute price	ATR	Only with other price-unit measures
Percentage (%)	Rate of change of range	Chaikin Volatility	No — a different quantity entirely

Because the log-return estimators share a scale, you can swap one for another, average two together, or annualise any of them with a single call. ATR stays in price units so it remains directly usable for stop distances; Chaikin measures whether the range is expanding or contracting, not its level.

The Estimators

The indicator exposes eight estimator options. The four range-based methods are the reason the toolkit exists; the classical measures are provided as baselines and for stop-placement work.

Estimator	Year	Reads	Scale	Character
Close-to-Close	—	Closing prices only	Log-return	Classical benchmark; least efficient. Also selectable as Standard Deviation (the same standard deviation of log returns).
Parkinson	1980	High–low range	Log-return	~5× efficient. Assumes zero drift; biased in trends.
Garman-Klass	1980	Open, high, low, close	Log-return	~7.4× efficient. Assumes zero drift and no gaps.
Rogers-Satchell	1991	OHLC with drift independence	Log-return	Unbiased under any drift; ideal for trending or 24/7 markets.
Yang-Zhang	2000	Overnight gap, open-close, and range	Log-return	Up to 14× efficient; handles drift and gaps. Uses internal rolling variance.
ATR	1978	True range, including gaps	Price ($)	Wilder’s familiar measure; directly usable for stops.
Chaikin	—	Rate of change of range	Percentage (%)	Expansion/contraction oscillator, not a level.

Choosing an estimator

• Trending equities with gaps — Yang-Zhang handles both drift and overnight jumps.
• Crypto (24/7) — Rogers-Satchell is drift-independent without Yang-Zhang’s multi-bar window.
• Ranging markets — Garman-Klass or Parkinson; simpler, no drift adjustment needed.
• Price-based stops — ATR, because its output is already in currency terms.

Smoothing Without the Drop-Off

Every estimator (except Yang-Zhang, which carries its own rolling variance) can be smoothed through ten filter types drawn from the FiltersToolkit Library: Ultimate Smoother (2/3-pole), Super Smoother (2/3-pole), BiQuad, ADXvma, MAMA, A2RMA, SMA, and EMA.

The choice matters more than it first appears. A simple moving average produces a drop-off artefact: when a volatile bar finally exits the averaging window, the reading crashes downward for no reason in the current market. Infinite Impulse Response filters such as the Ultimate Smoother and Super Smoother let old shocks decay gracefully instead, giving a volatility line you can actually reason about.

Note

The toolkit smooths variance for well-behaved filters but switches to smoothing volatility directly for overshooting filters, so that a filter which rings below zero on variance cannot produce an invalid negative reading.

Beyond a Single Line

The indicator is a full analysis surface, not just one plotted estimator.

Dual Estimator Slots

Run two estimators side by side, each with its own type, lookback, smoothing filter, and palette. Compare Yang-Zhang against Rogers-Satchell, or the same estimator under two different smoothers, directly on one chart.

Weighted Aggregation

Blend the two estimators into a single line with custom weights. The indicator normalises the weights automatically (so 2:1 becomes 67%:33%) and validates that both share a scale group, warning you if they do not.

Adaptive Thresholds

Detect regime shifts with three threshold modes: static (fixed values you set), dynamic (a baseline ± a standard-deviation multiplier), or percentile (the Nth percentile of recent history). Breaches are drawn with intensity-scaled fills that deepen the further volatility travels past the threshold.

Regime & Legend

An optional background tint marks HIGH, NORMAL, or LOW volatility — driven by Estimator 1, Estimator 2, or the aggregation. A legend table reports each estimator’s current value with its percentile rank (P##) for quick regime assessment.

Alerts cover the full workflow: estimator direction changes, an Estimator-1/Estimator-2 cross, high/low threshold breaches, and regime transitions — each available for both estimators and the aggregated line.

An Annualise Volatility toggle (on by default) expresses the log-return estimators in the familiar yearly convention, using a configurable periods-per-year value (252 for daily equities, 365 for crypto, and so on).

The Library, for Builders

If you are writing your own Pine scripts, the VolatilityToolkit Library exposes the whole workflow as reusable functions with standardised, per-bar log-return output — so swapping one estimator for another is a one-line change.

Function	Purpose
parkinson(), garman_klass(), rogers_satchell(), yang_zhang()	Range-based estimators (log-return units)
close_to_close()	Classical baseline (standard deviation of log returns)
atr()	Average True Range in price units, for stops and sizing
chaikin_volatility()	Range expansion/contraction oscillator
annualize()	Per-bar → yearly volatility via the √time rule
asymmetric_ewma()	Fast-up, slow-down smoothing — volatility spikes quickly, decays gently
weighted_horizon_volatility()	HAR-inspired blend across short, medium, and long horizons
volatility_mtf()	Multi-timeframe aggregation, scaling higher timeframes to the chart
volatility_burst_ratio()	Short-vs-long ratio for regime-shift and breakout detection

For the mathematically curious

Two helpers repay closer study. asymmetric_ewma() uses separate decay rates for rising and falling volatility (defaults alpha_up 0.1, alpha_down 0.02) — exactly the behaviour a trailing stop wants, widening fast on a shock and narrowing slowly afterwards. volatility_burst_ratio() compares short-term to long-term volatility: a ratio above 1.0 marks expansion, and readings above 1.5 often accompany breakouts. Note that weighted_horizon_volatility() is a HAR-inspired trailing average, not a forecasting regression — an honest distinction the library documents rather than glosses over.

Where It Fits in the Suite

The Volatility Toolkit is the shared measurement layer beneath our risk tooling. The Kinetic Trend Envelope draws five of these estimators — ATR, Parkinson, Garman-Klass, Rogers-Satchell, and Yang-Zhang — to size its asymmetric trailing-stop bands, calibrating them onto a comparable log-return basis so a single volatility factor behaves consistently whichever estimator you select. Reading this page explains where those five come from and how the remaining measures extend the same foundation.

Getting Started

1. Add the indicator

   Search for “GYTS Volatility Toolkit” in TradingView’s indicator library and add it to any chart.

2. Start with Yang-Zhang

   Estimator 1 defaults to Yang-Zhang (lookback 14) — the most comprehensive estimator for gapped markets. Keep Annualise Volatility enabled to read values in the standard yearly form.

3. Read the legend table

   Watch the current value and its percentile rank (P##). A reading such as “0.1523 (P75)” means 0.1523 annualised volatility, sitting at the 75th percentile of recent history.

4. Add a second view

   Enable Estimator 2 with a contrasting method — Rogers-Satchell is a natural counterpart — to see how different assumptions read the same market.

5. Turn on regime detection

   Enable thresholds (dynamic or percentile) and the regime background to let the toolkit flag elevated and suppressed states automatically.

Open Source Commitment

By publishing both the indicator and its library, we invite you to study the implementations, verify the mathematics against the original papers, and build derivative work of your own.

Source Code:

• Volatility Toolkit Indicator — the full interactive implementation
• VolatilityToolkit Library — reusable estimator, smoothing, and aggregation functions
• FiltersToolkit Library — the smoothing filters this toolkit builds upon

References

Source	Reference
Wilder	J. Welles Wilder Jr., New Concepts in Technical Trading Systems, 1978
Parkinson	M. Parkinson, “The Extreme Value Method for Estimating the Variance of the Rate of Return”, 1980, Journal of Business — DOI
Garman-Klass	M. B. Garman and M. J. Klass, “On the Estimation of Security Price Volatilities from Historical Data”, 1980, Journal of Business — DOI
Rogers-Satchell	L. C. G. Rogers and S. E. Satchell, “Estimating Variance from High, Low and Closing Prices”, 1991, Annals of Applied Probability — DOI
Yang-Zhang	D. Yang and Q. Zhang, “Drift-Independent Volatility Estimation Based on High, Low, Open, and Close Prices”, 2000, Journal of Business — DOI
Corsi	F. Corsi, “A Simple Approximate Long-Memory Model of Realized Volatility”, 2009, Journal of Financial Econometrics

Next Steps

• Explore the Filters Toolkit that supplies the smoothing options
• See how the Kinetic Trend Envelope turns these estimators into a trailing stop
• Study the Ultimate Smoother behind the drop-off-free smoothing
• Join our Discord to compare estimator behaviour across markets

Join Our Trading Community

Connect with fellow traders, share insights, and learn together.

Join Discord