Mean Reversion Quantitative Methodology - Technosight Documentation

1

Strategy Objective, Rationale & Market Regimes

Investment Objective: The Mean Reversion strategy identifies statistical overextensions in equity prices away from their historical central moving averages, generating counter-trend positions that capture profits as prices revert to equilibrium.

Foundational Economic & Behavioral Rationale: Rooted in the Ornstein-Uhlenbeck stochastic mean-reverting process and empirical studies by Fama and French (1988), Poterba and Summers (1988), and Lo and MacKinlay (1990). Market participants frequently overreact to short-term news, quarterly guidance revisions, or temporary liquidity imbalances, pushing prices beyond intrinsic statistical boundaries.

Market Regime Suitability:

• Optimal Regimes

Range-bound, sideways, choppy, and low-trend consolidating market environments with well-defined support/resistance bounds.

• Challenging Regimes

Strong secular bull runs or catastrophic structural breakdowns where prices trend persistently without mean reversion.

2

Market Data Inputs

The quantitative engine processes high-fidelity financial market data stored in high-performance QuestDB time-series tables across dual resolution layers:

1. End-of-Day (EOD) OHLCV Dataset (Primary)

Primary data source utilized for indicator calculations, mathematical factor modeling, and primary trade signal generation.

• Open, High, Low, Close (OHLC): Split & dividend adjusted.
• Volume: Total daily traded share volume.
• Universe Coverage: Active US equities spanning Russell 3000 and S&P 500.

2. 15-Minute Intraday Bar Dataset (Auxiliary)

High-frequency intraday bars utilized for auxiliary multi-timeframe confirmation and higher-timeframe regime alignment.

• 15m Interval Bars: Intraday pricing sequence.
• EMA(50, 200, 800) 15m: Triple Exponential Moving Average pattern detection.
• Purpose: Auxiliary signal evaluation and execution timing filter.

3

Indicators and Mathematical Formulations

The Mean Reversion model evaluates rolling statistical standard normal Z-scores to gauge distance from historical equilibrium:

$$\mu_K(t) = \frac{1}{K} \sum_{i=0}^{K-1} \text{Close}(t - i)$$

$$\sigma_K(t) = \sqrt{\frac{1}{K-1} \sum_{i=0}^{K-1} \left(\text{Close}(t - i) - \mu_K(t)\right)^2}$$

$$Z(t) = \frac{\text{Close}(t) - \mu_K(t)}{\sigma_K(t) + \epsilon}$$

$$\text{RawScore}(t) = -Z(t)$$

Where $K = 20$ trading days is the standard rolling lookback window.

4

Entry & Exit Rules (Trade Execution Logic)

The systematic trade logic for Mean Reversion triggers counter-trend entries at statistical extremes:

• Long Entry Trigger (+1.0 Signal - Oversold Reversion)

Condition: Rolling Z-score satisfies $Z(t) \le -\text{entry\_threshold}$ (default $Z \le -2.0$). Price is $\ge 2.0$ standard deviations below its 20-day mean.

• Short Entry Trigger (-1.0 Signal - Overbought Reversion)

Condition: Rolling Z-score satisfies $Z(t) \ge +\text{entry\_threshold}$ (default $Z \ge +2.0$). Price is $\ge 2.0$ standard deviations above its 20-day mean.

• Exit / Profit Target Trigger (0.0 Signal)

Condition: $|Z(t)| \le \text{exit\_threshold}$ (default $|Z| \le 0.5$). Position is closed once price returns within 0.5 standard deviations of its mean.