Wilder Moving Average (RMA)
The Wilder moving average, or RMA, is J. Welles Wilder's smoothing method — an exponential average with a gentler decay than a standard EMA — and it is the smoothing running inside RSI, ATR and ADX.
Wilder computed his smoothing by hand in the 1970s: take the previous average, subtract one Nth of it, add one Nth of the new value. That makes the RMA mathematically an EMA with a smaller smoothing factor — a 14-period RMA decays like a 27-period EMA — so it is slower and steadier than an EMA of the same stated length. It exists in this catalog at 14 periods because that is the length Wilder built his other indicators around.
Its main practical value is transparency into those other indicators. RSI is the RMA of gains against the RMA of losses; ATR is the RMA of true range; ADX smooths its directional components the same way. If an RSI or ATR reading surprises you, the RMA is the mechanism to check — its long memory is why RSI recovers from an extreme gradually rather than snapping back, and why ATR deflates slowly after a volatile spell.
Used directly as a trend line, the 14-period RMA behaves like a slow EMA: it holds its slope through noise that turns faster averages, and it arrives at genuine turns late. That deliberateness is a stance, not an oversight — Wilder wanted indicators that ignored single wild sessions — but it means an RMA cross is confirmation of a move already underway, never early warning of one.
Against the EMA the only difference is the decay rate, and it matters more than it looks: at the same stated period the RMA carries roughly twice the memory. If you want an average that behaves the way the internals of RSI and ATR behave — because you are screening on those and want the smoothing to match — this is the one.
In Sift
Written as rma(14). A working scan — stocks above Wilder's 14-period average with RSI agreeing above its midline:
where close > rma(14) and rsi(14) > 501
of the 500 most-traded NSE stocks match today, as of 20 Aug 2026
Common questions
What is the difference between RMA and EMA?
Only the smoothing factor. An EMA weights the newest bar by 2/(N+1); Wilder's RMA weights it by 1/N. At 14 periods that makes the RMA decay like a 27-period EMA — roughly twice the memory, half the nervousness. If a platform's RSI or ATR disagrees with yours, this substitution is the usual culprit.
Why does RSI use the Wilder moving average?
Because Wilder built both. He designed the RMA to be computed by hand with one addition and one subtraction a day, and its long memory suited his goal of indicators that ignore single-session noise. RSI, ATR and ADX all smooth their inputs this way, which is why they respond gradually rather than snapping to each new bar.