Quantitative Analysis: Momentum and Mean-Reversion

What is quantitative analysis?

Quantitative analysis (quant) uses statistical models and historical price data to detect patterns invisible to the naked eye. No gut feeling — pure mathematics. The two main quant strategies in the 360° system are momentum and mean-reversion.


Strategy 1: Momentum

Core thesis: What has risen strongly recently tends to keep rising — at least in the short term.

Momentum models measure how strongly a stock has performed relative to comparable stocks over a defined period (e.g. 20 or 60 trading days).

Momentum score Signal
Top 20 % of all candidates Strong long signal
Middle 60 % Neutral
Bottom 20 % Short signal

Momentum works best in trending conditions and fails in sideways markets.


Strategy 2: Mean-Reversion

Core thesis: Extreme price moves revert to the mean — an exaggerated drop is often corrected.

The Z-score measures how far the current price deviates from its historical average (in standard deviations).

Z-score Interpretation Signal
< −2.0 Strongly oversold Long (reversion expected)
−2.0 to −1.0 Slightly oversold Mild long bias
−1.0 to +1.0 Normal Neutral
+1.0 to +2.0 Slightly overbought Mild short bias
> +2.0 Strongly overbought Short (reversion expected)

Mean-reversion works best in sideways markets and for stocks without a strong trend.


Combination: Momentum vs. Mean-Reversion

The quant score combines both strategies — weighted by the current market regime:

Bull regime    → Momentum weight 70 %, mean-reversion 30 %
Bear regime    → Momentum weight 60 %, mean-reversion 40 %
Sideways       → Momentum weight 40 %, mean-reversion 60 %

The market regime (→ market regime analysis) therefore influences how the quant score is calculated.


Additional quant metrics in the 360° system

  • Sharpe ratio (rolling 30 days): Return relative to risk
  • Beta: How much does the stock fluctuate relative to the S&P 500?
  • Correlation to SPY: Does the stock move with the market or independently?

Limitations of quantitative analysis

  • Models are based on historical data — past patterns do not guarantee the future
  • Quant signals can be instantly invalidated by news or earnings
  • Overfitting risk: models that work perfectly on old data fail on new data
  • Candidates from the selection universe often have short trading histories (frequently small-caps) — statistical significance is limited

All 8 methods → | Market regime → | Liquidity analysis → | Pricing →