> ## Documentation Index
> Fetch the complete documentation index at: https://docs.chicago.global/llms.txt
> Use this file to discover all available pages before exploring further.

# Scoring Methodology

> How Parallax calculates factor scores, integrates them with ML optimization, and generates investment recommendations

Our scoring system transforms complex financial data into clear, actionable investment insights. Built on Fama-French factor research and other academic foundations, our models incorporate real-time data and contemporary measures.

## Score Calculation Process

### Individual Factor Scores (0-10 Scale)

Scores represent **rounded percentiles** within the investment universe:

| Score Range | Percentile      | Interpretation                    |
| ----------- | --------------- | --------------------------------- |
| **8-10**    | Top quartile    | Strong to exceptional performance |
| **6-7**     | Above median    | Moderately attractive             |
| **4-5**     | Below median    | Neutral to weak                   |
| **0-3**     | Bottom quartile | Poor to very poor performance     |

<Note>
  **Percentile Interpretation**: A Value score of 10 means the stock is in the **top 10% cheapest** relative to fundamentals. A Quality score of 3 means it's in the **bottom 20%** for business quality metrics.
</Note>

### Overall Parallax Score Integration

The Overall Score uses a **machine learning model** that dynamically adapts to changing market environments:

```
Overall Score = ML-Optimized Combination of:
  Value Factor     (10-40% dynamic weight)
  Quality Factor   (10-40% dynamic weight)
  Momentum Factor  (10-40% dynamic weight)
  Defensive Factor (10-40% dynamic weight)
  Tactical Factor  (10-40% dynamic weight)

Note: Size is NOT included in Overall Score calculation
```

## ML Optimization Process

<Steps>
  <Step title="Predict Returns">
    Models forecast expected returns for each factor based on current market regime
  </Step>

  <Step title="Estimate Covariances">
    Predicts how factors will correlate in the near-term
  </Step>

  <Step title="Optimize Weights">
    Allocates to factors with best risk-adjusted return forecasts
  </Step>

  <Step title="Enforce Limits">
    Each factor weight constrained between 10% (minimum) and 40% (maximum)
  </Step>

  <Step title="Periodic Updates">
    Weights recalculated periodically as market conditions evolve
  </Step>
</Steps>

<Tip>
  **Example**: During high-volatility periods, the ML model might increase Defensive factor weight to 35% and reduce Momentum to 15%, automatically adapting your strategy to market conditions.
</Tip>

## Size Factor Treatment

The Size factor has a unique role in the framework:

| Aspect                     | Treatment                                                     |
| -------------------------- | ------------------------------------------------------------- |
| **Overall Score**          | Size does NOT contribute to individual stock recommendations  |
| **Portfolio Construction** | Size is systematically incorporated during portfolio building |
| **Liquidity Management**   | Ensures adequate trading capacity and minimizes market impact |
| **Capacity Constraints**   | Helps manage strategy scalability and position sizing         |
| **Factor Amplification**   | Small-cap exposure can amplify other factor signals           |

## Recommendation Generation

Overall scores translate directly into investment recommendations:

| Score Range  | Recommendation |
| ------------ | -------------- |
| **8.5-10.0** | STRONG BUY     |
| **6.5-8.4**  | BUY            |
| **3.5-6.4**  | HOLD           |
| **1.5-3.4**  | SELL           |
| **0.0-1.4**  | STRONG SELL    |

## Score Validation and Quality Control

### Multi-Source Verification

* Cross-validation across fundamental, technical, and alternative data
* Peer comparison within sector and market cap categories
* Historical score performance tracking
* Outlier detection and manual review processes

### Dynamic Adjustment via ML Model

| Adjustment Type                 | Description                                                              |
| ------------------------------- | ------------------------------------------------------------------------ |
| **Market Regime Detection**     | Identifies changing market conditions (bull, bear, high-vol, low-vol)    |
| **Adaptive Factor Weights**     | ML model adjusts Overall Score weights periodically within 10-40% bounds |
| **Covariance Forecasting**      | Predicts how factors will correlate in current regime                    |
| **Return Prediction**           | Estimates near-term factor performance based on market environment       |
| **Sector-Specific Adjustments** | Scoring refined for industry-specific characteristics                    |
| **Event-Driven Modifications**  | Earnings season and corporate action adjustments                         |

## Performance Attribution

Understanding where returns come from through factor decomposition:

```
Example 6-Month Performance Attribution:
Total Return: +8.3%
  Market Beta (1.1):      +5.2%
  Value Factor (0.4):     -0.8%
  Momentum Factor (0.6):  +2.1%
  Quality Factor (0.5):   +1.4%
  Size Factor (-0.2):     -0.3%
  Tactical Factor (0.2):  +0.4%
  Security Selection:     +0.3%
```

### Factor Risk Analysis

Understanding portfolio risk sources through factor decomposition:

* **Systematic Risk**: Market and factor exposures
* **Specific Risk**: Individual security risk
* **Concentration Risk**: Factor and sector concentrations
* **Correlation Risk**: How holdings relate during stress periods

## Continuous Improvement Process

<CardGroup cols={3}>
  <Card title="Monthly Reviews" icon="calendar">
    Individual factor performance analysis, interaction assessment, and scoring accuracy validation
  </Card>

  <Card title="Quarterly Updates" icon="arrows-rotate">
    Factor weight optimization, new data source integration, and model refinement
  </Card>

  <Card title="Annual Review" icon="clipboard-check">
    Complete framework assessment, academic research updates, and technology enhancements
  </Card>
</CardGroup>