Concept
EWMA Volatility
EWMA Volatility, also known as RiskMetrics λ-decay, is a Volatility concept. The Library holds 2 implementations, each one a working definition you can pull into Quant.
Top EWMA Volatility indicators
2 total
What is EWMA Volatility?
EWMA volatility estimates variance with an exponentially weighted moving average of squared returns: today's estimate equals a decay factor lambda times yesterday's estimate, plus one minus lambda times yesterday's squared return, with the mean return usually set to zero. J.P. Morgan's RiskMetrics methodology popularized the recursion and fixed lambda at 0.94 for daily data, which is why the recursion is often just called the RiskMetrics model. Geometrically decaying weights let the freshest shocks dominate while old ones fade smoothly.
Two properties drive its use. It has no drop-off artifact: rolling close-to-close historical volatility jumps when a large return enters the window and again when it exits, while EWMA decays a shock continuously. And it is the boundary case of the GARCH family with no long-run anchor: it captures volatility clustering but never mean-reverts, so its forecast at any horizon is simply the current estimate.
How traders use it
- As the volatility input for risk work: volatility-targeted sizing, value-at-risk style estimates, and volatility-scaled stops benefit from an estimate that updates smoothly every bar.
- As an adaptive width for bands and filters: envelopes built on EWMA variance widen quickly after shocks and tighten as calm returns, without the step changes a fixed rolling window produces.
- As a freshness read: EWMA volatility above a slower equal-weighted estimate says turbulence is recent and building; below it, decaying.
Related concepts · Volatility estimators
Concept family
Volatility
56 concepts mapped · 43 in the Library
EWMA Volatility FAQ
What lambda should I use for EWMA volatility?
RiskMetrics' classic values are 0.94 for daily returns and 0.97 for monthly, and most implementations default near those. Higher lambda decays slower, giving a smoother, steadier estimate; lower lambda tracks shocks faster at the cost of noise. There is no universally correct value: it is a responsiveness-versus-stability trade, so match it to your holding horizon.
How is EWMA volatility different from GARCH?
EWMA is the special case of GARCH(1,1) with the persistence terms summing to one and no constant, so there is no long-run variance to revert toward. GARCH must be fitted but forecasts volatility decaying back to its unconditional level; EWMA needs no estimation and forecasts a flat line at today's value, which makes it the simpler choice for bar-by-bar chart calculations.
Build EWMA Volatility your way.
Quant writes, tests, and refines it with you — then it runs on LuxAlgo charting or ports to TradingView.