Quantitative Definitions of Momentum Beyond Moving Averages

Summary

The document clarifies that momentum strategies can be quantitative even when they use moving averages. It notes that moving average approaches range from simple and exponential forms to calculations on irregularly spaced tick data, as well as models that combine several averages with weights that change over time.

It also distinguishes this technical-indicator framing from the academic use of momentum, which commonly ranks assets by relative performance over a lookback period, sometimes after adjusting for a risk-factor model. The term can therefore refer to different signal definitions, and moving averages are only one possible approach. The document explains the distinction but does not specify a complete trading rule, test results, or evidence about which momentum measure performs best; those choices require further research and validation.

Key ideas

• Moving averages are quantitative signals and can be built in multiple ways.
• Momentum models can combine several moving averages with changing weights.
• Academic momentum commonly ranks assets by past relative performance over a chosen period.
• Factor-adjusted relative performance is distinct from a moving average signal.

Tags

# Can momentum strategies be quantitative in nature?


# Can momentum strategies be quantitative in nature?












I have read some papers on quantitative trading strategies and it seems like strategies that focus on mean reversion or statistical arbitrage give signals that are dependent on some quantitative model.

But when you turn to momentum strategies, people start talking about moving averages to generate signals. Why is that? Isn't there a more quantitative method of detecting momentum than moving averages ?

## Answer by Tal Fishman (score 6, accepted)

https://quant.stackexchange.com/a/2562

As jk3000 writes in his comment, moving averages are quantitative. Moving averages can be made quite sophisticated, if desired. Besides simple moving averages, there are exponential moving averages, moving averages on inhomogenous (tick) data, and meta models incorporating multiple moving averages with time-varying weights.

Also, the canonical academic definition of momentum (Jegadeesh and Titman (1993)) looks at top/bottom performers over some time period, often relative to a model such as CAPM or Fama-French. Momentum can mean different things to different people, but to serious quants, it is more often something along the lines of relative performance than moving averages. See http://momentum.behaviouralfinance.net/ to get a better idea.

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.