Introducing Slow Momentum with Fast Reversion for after-work quants

Takeaway. Markets can show slow trend-following behaviour and fast mean-reversion at once—so a single horizon rarely tells the whole story; combining slow momentum with fast reversion is a modern multi-scale idea.

Slow Momentum with Fast Reversion (Kieran Wood, Stephen Roberts, Stefan Zohren, 2021) sits in the momentum-ML lane after-work learners meet after classic Jegadeesh–Titman.

Fit for learners: Combines slow trend signals with fast mean-reversion overlays—modern momentum-ML intuition for multi-horizon features.

This Hub note is a plain-language intro. Not investment advice. Pair with Jegadeesh–Titman and spatio-temporal momentum.

The problem

A lone 12-month momentum score ignores that short horizons often reverse while longer horizons trend. Learners who hard-code one lookback miss the multi-scale structure papers in this line try to capture.

The idea / core intuition

• Separate slow momentum (longer lookbacks / slower signals) from fast reversion components.
• Show how blending scales can improve the story versus a single-horizon baseline (see paper for methods and markets).
• Think in features across horizons, not one magic window.
• Still research—costs, regime breaks, and overfitting remain first-class risks.

Why after-work Python learners care

This paper is a mental model for feature engineering: stack returns at 1d/5d/21d/63d-style windows, then ask what is trend vs noise. Perfect once you can rolling in pandas.

Tiny evening exercise: on one toy price series, compute slow (e.g. 60-day) and fast (e.g. 5-day) return signs; count days they agree vs disagree—learning only.

One-evening reading checklist

1. Contrast slow momentum vs fast reversion in plain words.
2. Explain why one lookback is often too blunt.
3. Note what the paper studies (asset class / method family) from the abstract.
4. List one overfitting trap when stacking many horizons.

Download PDF

Download PDF

Landing: arXiv abs

Important disclaimer

Not investment advice. This Hub note is educational only. The paper is research on multi-horizon signals—not a live trading system.

What this is not

• Not CEF / subsidy framing
• Not Quant or Macro course CTA
• Not a CTA-product advertisement

Next step

Single course CTA: SoloQuant Academy’s Introduction to Python Programming—master rolling windows and plots first; then multi-horizon feature notebooks get much easier.