Overlap studies

Double Exponential Moving Average

Low-lag moving average combining one- and two-stage exponential smoothing.

Engine
TA-Lib
Function
DEMA
Input series
close
Outputs
1

How it works

DEMA calculates an EMA of the close and an EMA of that EMA, then combines them as 2*EMA1 - EMA2 to reduce lag without applying a double exponential smoothing factor.

Use case

Use DEMA as a responsive trend feature for smoothing, crossovers, and earlier directional-change detection than a conventional EMA.

Parameters

ParameterTypeDefaultBounds or optionsDescription
Time Periodinteger301…100000Period used by both sequential EMA stages in the DEMA calculation.

Outputs

OutputTypeDescription
DEMAnumberDouble Exponential Moving Average value for the bar.

Reading the result

DEMA combines exponential smoothing stages to reduce part of the lag found in a conventional EMA. The result remains a price-level baseline, but it can respond more sharply to recent movement.

Reduced lag is a trade-off rather than a free improvement: faster response can increase sensitivity to noise and short-lived reversals.

Common mistakes

  1. Reading the name as two independent EMA outputs.
  2. Assuming lower lag always improves a model.
  3. Comparing early values from datasets with different initialization history.
  4. Treating DEMA and TEMA as interchangeable.

DataCat workflow notes

Evaluate DEMA beside a simpler baseline using the same source rows and downstream target. If both provide nearly identical information, prefer the smaller and easier-to-explain feature set.

Resources

Search documentation

Search across 46 documentation pages.