Skip to content
    Flow-alerta.cab

    Exponential Moving Average (EMA)

    AI fundamentals & modelsExponential moving averageEMAexponentially weighted moving average
    Editorially reviewed on: 2 sources
    Your reading view

    The short answer · 11 words

    In short

    EMA smoothes a time series and weights recent observations more strongly.

    Flow-alerta.cab (2026): Exponential Moving Average (EMA). https://flow-alerta.cab/en/glossar/exponential-moving-average-ema. Editorially reviewed on: 10 October 2026.

    How much detail would you like?

    Practical

    EMA supports the smoothing of noisy key figures. For alarms and interpretation, you should consider the raw values and the resulting delay.

    Hypothetical example: A dashboard smooths daily ticket numbers without completely hiding short-term changes.

    SMEs can visualise noisy operating indicators, should raw values and alarm rules be given.

    At a glance

    Core
    In the usual convention, EMA_t = α · x_t + (1 − α) · EMA_(t−1), with a smoothing factor between 0 and 1. A larger α weights the new value more strongly; initialization and convention used must be established.
    Use
    EMA supports the smoothing of noisy key figures. For alarms and interpretation, you should consider the raw values and the resulting delay.
    Example
    Hypothetical example: A dashboard smooths daily ticket numbers without completely hiding short-term changes.
    Limits
    Choose smoothing factor without target; output smoothed series as prognosis; ignore delay and outliers.

    In context

    Exponential Smoothing

    Exponential smoothing is a family of time series methods with decreasing weights for older observations.

    Questions about this term

    How does a high smoothing factor react?

    Faster to new values.

    Is EMA a prognosis?

    Primarily a smoothing.

    All details and background

    In the usual convention, EMA_t = α · x_t + (1 − α) · EMA_(t−1), with a smoothing factor between 0 and 1. A larger α weights the new value more strongly; initialization and convention used must be established.

    ELI5 – Explained simply

    New values count more, older ones have an increasingly weaker effect.

    Why it matters in practice

    EMA supports the smoothing of noisy key figures. For alarms and interpretation, you should consider the raw values and the resulting delay.

    Example

    Hypothetical example: A dashboard smooths daily ticket numbers without completely hiding short-term changes.

    Pitfalls and limits

    • Choose smoothing factor without target; output smoothed series as prognosis; ignore delay and outliers.

    Origin & history

    The recursive mean corresponds to the level update of simple exponential smoothing; this method was described in the late 1950s.

    Use in mid-sized companies

    • SMEs can visualise noisy operating indicators, should raw values and alarm rules be given.

    Related terms

    Matching Flow-alerta service

    AI opportunity analysis & roadmapAI consulting: Compare suitable smoothing and alarm rules for a process key figure.

    Matching tool / use case

    View use casesUse Case: smoothed process parameter next to raw data

    Sources

    Quellenprüfung: 10.10.2026. Definition und dokumentierte Einordnung abgeglichen; Beispiele beschreiben mögliche Anwendungen.

    Editorially reviewed on:

    Next step

    Put knowledge into practice

    Discuss your projectView use cases