Understanding Drivers

Educational Content: This documentation explains statistical relationships observed in historical data. It is not investment advice. Correlations are based on past data and do not predict future performance. All investments carry risk of loss.

Quick Start: Reading Your First Driver

Let's decode a real driver together:

Example Driver:

    
    insider_buying_20d → forward_return_30d
r=0.74, n=156, contribution=28.5%

In plain English: When insiders bought stock over the past 20 days, the stock price was higher 30 days later in 74% of historical cases. We've seen this pattern 156 times. This single pattern accounts for 28.5% of the total opportunity score.

Now let's break down each piece...

Understanding Historical Patterns

Drivers are statistical relationships observed between market events and subsequent price movements in historical data.

For example: when insider buying increased 200% in past observations, stocks showed an average 12% rise over the following 30 days.

The system identifies the top 5 strongest relationships for each setup, ranked by their statistical contribution.

What Are Drivers?

Drivers represent statistical correlations observed in historical market data. Here are examples:

Example 1: Insider Buying Pattern

WHEN insider buying increased 200%

Historical observation showed average 12% price increase over following 30 days

Example 2: Momentum Crossover

WHEN momentum crossed above 80

Historical data showed continued upward movement for 10-20 days on average

Example 3: Valuation Reversion

WHEN P/E dropped below sector average

Historical patterns showed mean reversion toward sector valuation levels

These relationships have been tested across hundreds of occurrences to establish statistical significance.

Anatomy of a Driver

    
    insider_buying_20d → forward_return_30d

Signal (what happened): Insider buying over the last 20 days
Outcome (what followed): Stock price 30 days later

The Three Critical Numbers

Every driver shows you three critical numbers:

1. Signal Strength (r-value)

How reliably these two things move together:

r-value	Interpretation
0.7 - 1.0	Very Strong
0.5 - 0.7	Strong
0.3 - 0.5	Moderate
Below 0.3	Weak (filtered out)

2. Times Tested (n)

Number of times we've verified this relationship:

Sample Size	Reliability
100+	High Reliability
50 - 100	Moderate
30 - 50	Use Caution

3. Contribution (%)

How much this driver influences the overall score.

A driver with 30% contribution is responsible for 30 points of a 100-point score. Higher contribution = more important to the opportunity.

Real Example: Reading a Driver

    
    insider_buying_20d → forward_return_30d
r=0.74, n=156, contribution=28.5%

Translation:

Signal Strength (r=0.74): This is a strong relationship. When insiders buy heavily, the stock rises 74% of the time.
Times Tested (n=156): We've verified this pattern 156 times. High reliability.
Contribution (28.5%): This single driver accounts for 28.5% of the overall opportunity score.

What This Means:

When company insiders have been buying shares over the past 20 days, we've observed that the stock price tends to be higher 30 days later. This has happened 156 times in our historical data, with a 74% correlation. This pattern is strong enough to contribute 28.5% to the total opportunity score.

Common Driver Patterns You'll See

Pattern 1: Momentum Confirmation

Multiple technical indicators all pointing in the same direction.

    
    MA crosses → forward_return_10d (r=+0.45, n=203)
MACD bullish → forward_return_10d (r=+0.42, n=187)
RSI strong → forward_return_10d (r=+0.38, n=156)

Interpretation: When all technical indicators agree the stock is bullish, the 10-day forward returns tend to be positive. High conviction signal.

Pattern 2: Mean Reversion

Negative correlation = "what goes up must come down"

    
    RSI overbought → forward_return_5d (r=-0.52, n=134)

Interpretation: The negative correlation means when RSI signals "overbought," short-term returns tend to be negative. The stock often pulls back. This is a fade signal, not a chase signal.

Pattern 3: Insider Edge

Following the smart money.

    
    net_buy_ratio_10d → forward_return_20d (r=+0.41, n=89)
insider_count_10d → forward_return_20d (r=+0.36, n=76)

Interpretation: When insiders are net buyers (more buying than selling), the stock tends to perform well over the next 20 days. Insiders know something the market doesn't.

Pattern 4: Volatility Regime Shift

Predicting wild swings vs calm periods.

    
    BB breakouts → realized_vol_10d (r=+0.82, n=145)

Interpretation: Bollinger Band breakouts strongly predict sustained volatility expansion. This is an options play, not a directional bet. Trade volatility, not direction.

Analyzing Driver Quality

Pattern Alignment

Examine whether the top 5 drivers show consistent directionality or conflicting signals.

Strong Alignment All correlations point same direction (all positive or all negative r-values)

Mixed Signals Correlations conflict (mix of positive and negative r-values)

Statistical Strength Indicators

Key metrics for evaluating driver reliability:

|r| ≥ 0.7 - Indicates very strong historical correlation
n ≥ 100 - Represents robust sample size for statistical validity
High contribution - Drivers contributing 20%+ have greater impact on overall score

Time Horizon Context

Drivers analyze different forward-looking timeframes in historical data:

Timeframe	Pattern Type
5-day forward returns	Short-term patterns
10-20 day forward returns	Medium-term patterns
30-day forward returns	Longer-term patterns