Definitions
In plain terms
A larger smoothing constant makes an EMA respond faster to new data, while a smaller value makes it change more slowly and remember older observations longer.
Technical
In EMA_t = EMA_(t-1) + alpha(z_t - EMA_(t-1)), alpha is a dimensionless value between zero and one that multiplies the current forecast error.
Scope
This concept describes the recurrence weight; it does not independently specify initialization, warm-up length, missing-data handling, or predictive value.
Examples
- Alpha 0.10 closes 10 percent of the gap between the prior EMA and the current observation.
Common misconceptions
- A 10 percent smoothing constant does not mean only the latest ten observations are used.
Concept relationships
Where this concept is used
Tutorials planned
These catalogued topics use this concept, but their complete build has not shipped yet.
- D04-F02-A01 Important
- D04-F02-A02 Important
- D04-F02-A03 Important
- D04-F02-A05 Important
- D04-F02-A06 Important
- D07-F01-A02 Important
- D07-F01-A04 Important
- D07-F01-A05 Prerequisite
- D07-F01-A06 Prerequisite
- D07-F01-A08 Important
- D07-F01-A09 Important
- D07-F02-A01 Prerequisite
Evidence and governance
- Calculating the McClellan Oscillator McClellan Financial Publications · first party methodology
Supports: preferred label, short definition, technical definition, formula, units
Limits: Does not prescribe one universal initialization, missing-session, revision-resolution, rounding, or software interface policy.
- Reviewed by
- Islam Baraka
- Last reviewed
- 2026-07-26
- Next review
- 2027-07-26
- Record status
- published
