Exponential Smoothing of Returns and Forecasting Limits

Summary

The note distinguishes exponential decay in a physical process from exponential smoothing used to weight financial returns. For a return forecast, recent observations receive greater weight, with the decay rate often expressed through a half-life parameter. The answer directs readers toward statistical treatments of exponential smoothing rather than simply substituting returns into a decay formula.

It provides no derivation, worked calculation, or empirical comparison. The respondent cautions that smoothing returns may not produce a useful forecast and argues that a straightforward method is unlikely to deliver a strong predictive edge. The note therefore offers a conceptual pointer, not a validated trading strategy; it does not specify how to select the half-life, assess forecast performance, or handle changing market conditions.

Key ideas

• Exponential smoothing weights recent returns more heavily to form a forecast.
• A half-life parameter controls how quickly the influence of older returns declines.
• A physical exponential-decay formula should not be transferred to returns without considering the statistical model.
• The answer questions whether smoothed returns will provide useful forecasting power and supplies no empirical evidence.

Tags

# Exponential weighting of returns


# Exponential weighting of returns












I am looking for a procedure to compute an exponential weighting of returns given a half life parameter.

I ran accross a wikipedia article, can I take it unchanged an assume N(t) is the return at time t ?

## Answer by Richi Wa (score 0, accepted)

https://quant.stackexchange.com/a/16918

What you want to do sounds like exponential smoothing of returns. So you want to forecast a return by exponentially weighting recent returns. For exponential smoothing you can look at Hyndman's papers and the e-book or this course on page 34.

I personally don't think that this will give a good forecast! It would be much too easy by the way.

The link that you provide points to exponential decay. We had this in school when radio-active material got less and less. I would not just replace $N(t)$ in the wikipedia article but rather read about statistics e.g. using Hyndman's ressources (I know I cite him a lot ...).

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.