OPEN-SOURCE SCRIPT
Advanced Petroleum Market Model (APMM)

Advanced Petroleum Market Model (APMM): A Multi-Factor Fundamental Analysis Framework for Oil Market Assessment
## 1. Introduction
The petroleum market represents one of the most complex and globally significant commodity markets, characterized by intricate supply-demand dynamics, geopolitical influences, and substantial price volatility (Hamilton, 2009). Traditional fundamental analysis approaches often struggle to synthesize the multitude of relevant indicators into actionable insights due to data heterogeneity, temporal misalignment, and subjective weighting schemes (Baumeister & Kilian, 2016).
The Advanced Petroleum Market Model addresses these limitations through a systematic, quantitative approach that integrates 16 verified fundamental indicators across five critical market dimensions. The model builds upon established financial engineering principles while incorporating petroleum-specific market dynamics and adaptive learning mechanisms.
## 2. Theoretical Framework
### 2.1 Market Efficiency and Information Integration
The model operates under the assumption of semi-strong market efficiency, where fundamental information is gradually incorporated into prices with varying degrees of lag (Fama, 1970). The petroleum market's unique characteristics, including storage costs, transportation constraints, and geopolitical risk premiums, create opportunities for fundamental analysis to provide predictive value (Kilian, 2009).
### 2.2 Multi-Factor Asset Pricing Theory
Drawing from Ross's (1976) Arbitrage Pricing Theory, the model treats petroleum prices as driven by multiple systematic risk factors. The five-factor decomposition (Supply, Inventory, Demand, Trade, Sentiment) represents economically meaningful sources of systematic risk in petroleum markets (Chen et al., 1986).
## 3. Methodology
### 3.1 Data Sources and Quality Framework
The model integrates 16 fundamental indicators sourced from verified TradingView economic data feeds:
Supply Indicators:
- US Oil Production (ECONOMICS:USCOP)
- US Oil Rigs Count (ECONOMICS:USCOR)
- API Crude Runs (ECONOMICS:USACR)
Inventory Indicators:
- US Crude Stock Changes (ECONOMICS:USCOSC)
- Cushing Stocks (ECONOMICS:USCCOS)
- API Crude Stocks (ECONOMICS:USCSC)
- API Gasoline Stocks (ECONOMICS:USGS)
- API Distillate Stocks (ECONOMICS:USDS)
Demand Indicators:
- Refinery Crude Runs (ECONOMICS:USRCR)
- Gasoline Production (ECONOMICS:USGPRO)
- Distillate Production (ECONOMICS:USDFP)
- Industrial Production Index (FRED:INDPRO)
Trade Indicators:
- US Crude Imports (ECONOMICS:USCOI)
- US Oil Exports (ECONOMICS:USOE)
- API Crude Imports (ECONOMICS:USCI)
- Dollar Index (TVC:DXY)
Sentiment Indicators:
- Oil Volatility Index (CBOE:OVX)
### 3.2 Data Quality Monitoring System
Following best practices in quantitative finance (Lopez de Prado, 2018), the model implements comprehensive data quality monitoring:
Data Quality Score = Σ(Individual Indicator Validity) / Total Indicators
Where validity is determined by:
- Non-null data availability
- Positive value validation
- Temporal consistency checks
### 3.3 Statistical Normalization Framework
#### 3.3.1 Z-Score Normalization
The model employs robust Z-score normalization as established by Sharpe (1994) for cross-indicator comparability:
Z_i,t = (X_i,t - μ_i) / σ_i
Where:
- X_i,t = Raw value of indicator i at time t
- μ_i = Sample mean of indicator i
- σ_i = Sample standard deviation of indicator i
Z-scores are capped at ±3 to mitigate outlier influence (Tukey, 1977).
#### 3.3.2 Percentile Rank Transformation
For intuitive interpretation, Z-scores are converted to percentile ranks following the methodology of Conover (1999):
Percentile_Rank = (Number of values < current_value) / Total_observations × 100
### 3.4 Exponential Smoothing Framework
Signal smoothing employs exponential weighted moving averages (Brown, 1963) with adaptive alpha parameter:
S_t = α × X_t + (1-α) × S_{t-1}
Where α = 2/(N+1) and N represents the smoothing period.
### 3.5 Dynamic Threshold Optimization
The model implements adaptive thresholds using Bollinger Band methodology (Bollinger, 1992):
Dynamic_Threshold = μ ± (k × σ)
Where k is the threshold multiplier adjusted for market volatility regime.
### 3.6 Composite Score Calculation
The fundamental score integrates component scores through weighted averaging:
Fundamental_Score = Σ(w_i × Score_i × Quality_i)
Where:
- w_i = Normalized component weight
- Score_i = Component fundamental score
- Quality_i = Data quality adjustment factor
## 4. Implementation Architecture
### 4.1 Adaptive Parameter Framework
The model incorporates regime-specific adjustments based on market volatility:
Volatility_Regime = σ_price / μ_price × 100
High volatility regimes (>25%) trigger enhanced weighting for inventory and sentiment components, reflecting increased market sensitivity to supply disruptions and psychological factors.
### 4.2 Data Synchronization Protocol
Given varying publication frequencies (daily, weekly, monthly), the model employs forward-fill synchronization to maintain temporal alignment across all indicators.
### 4.3 Quality-Adjusted Scoring
Component scores are adjusted for data quality to prevent degraded inputs from contaminating the composite signal:
Adjusted_Score = Raw_Score × Quality_Factor + 50 × (1 - Quality_Factor)
This formulation ensures that poor-quality data reverts toward neutral (50) rather than contributing noise.
## 5. Usage Guidelines and Best Practices
### 5.1 Configuration Recommendations
For Short-term Analysis (1-4 weeks):
- Lookback Period: 26 weeks
- Smoothing Length: 3-5 periods
- Confidence Period: 13 weeks
- Increase inventory and sentiment weights
For Medium-term Analysis (1-3 months):
- Lookback Period: 52 weeks
- Smoothing Length: 5-8 periods
- Confidence Period: 26 weeks
- Balanced component weights
For Long-term Analysis (3+ months):
- Lookback Period: 104 weeks
- Smoothing Length: 8-12 periods
- Confidence Period: 52 weeks
- Increase supply and demand weights
### 5.2 Signal Interpretation Framework
Bullish Signals (Score > 70):
- Fundamental conditions favor price appreciation
- Consider long positions or reduced short exposure
- Monitor for trend confirmation across multiple timeframes
Bearish Signals (Score < 30):
- Fundamental conditions suggest price weakness
- Consider short positions or reduced long exposure
- Evaluate downside protection strategies
Neutral Range (30-70):
- Mixed fundamental environment
- Favor range-bound or volatility strategies
- Wait for clearer directional signals
### 5.3 Risk Management Considerations
1. Data Quality Monitoring: Continuously monitor the data quality dashboard. Scores below 75% warrant increased caution.
2. Regime Awareness: Adjust position sizing based on volatility regime indicators. High volatility periods require reduced exposure.
3. Correlation Analysis: Monitor correlation with crude oil prices to validate model effectiveness.
4. Fundamental-Technical Divergence: Pay attention when fundamental signals diverge from technical indicators, as this may signal regime changes.
### 5.4 Alert System Optimization
Configure alerts conservatively to avoid false signals:
- Set alert threshold at 75+ for high-confidence signals
- Enable data quality warnings to maintain system integrity
- Use trend reversal alerts for early regime change detection
## 6. Model Validation and Performance Metrics
### 6.1 Statistical Validation
The model's statistical robustness is ensured through:
- Out-of-sample testing protocols
- Rolling window validation
- Bootstrap confidence intervals
- Regime-specific performance analysis
### 6.2 Economic Validation
Fundamental accuracy is validated against:
- Energy Information Administration (EIA) official reports
- International Energy Agency (IEA) market assessments
- Commercial inventory data verification
## 7. Limitations and Considerations
### 7.1 Model Limitations
1. Data Dependency: Model performance is contingent on data availability and quality from external sources.
2. US Market Focus: Primary data sources are US-centric, potentially limiting global applicability.
3. Lag Effects: Some fundamental indicators exhibit publication lags that may delay signal generation.
4. Regime Shifts: Structural market changes may require model recalibration.
### 7.2 Market Environment Considerations
The model is optimized for normal market conditions. During extreme events (e.g., geopolitical crises, pandemics), additional qualitative factors should be considered alongside quantitative signals.
## References
Baumeister, C., & Kilian, L. (2016). Forty years of oil price fluctuations: Why the price of oil may still surprise us. *Journal of Economic Perspectives*, 30(1), 139-160.
Bollinger, J. (1992). *Bollinger on Bollinger Bands*. McGraw-Hill.
Brown, R. G. (1963). *Smoothing, Forecasting and Prediction of Discrete Time Series*. Prentice-Hall.
Chen, N. F., Roll, R., & Ross, S. A. (1986). Economic forces and the stock market. *Journal of Business*, 59(3), 383-403.
Conover, W. J. (1999). *Practical Nonparametric Statistics* (3rd ed.). John Wiley & Sons.
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. *Journal of Finance*, 25(2), 383-417.
Hamilton, J. D. (2009). Understanding crude oil prices. *Energy Journal*, 30(2), 179-206.
Kilian, L. (2009). Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market. *American Economic Review*, 99(3), 1053-1069.
Lopez de Prado, M. (2018). *Advances in Financial Machine Learning*. John Wiley & Sons.
Ross, S. A. (1976). The arbitrage theory of capital asset pricing. *Journal of Economic Theory*, 13(3), 341-360.
Sharpe, W. F. (1994). The Sharpe ratio. *Journal of Portfolio Management*, 21(1), 49-58.
Tukey, J. W. (1977). *Exploratory Data Analysis*. Addison-Wesley.
## 1. Introduction
The petroleum market represents one of the most complex and globally significant commodity markets, characterized by intricate supply-demand dynamics, geopolitical influences, and substantial price volatility (Hamilton, 2009). Traditional fundamental analysis approaches often struggle to synthesize the multitude of relevant indicators into actionable insights due to data heterogeneity, temporal misalignment, and subjective weighting schemes (Baumeister & Kilian, 2016).
The Advanced Petroleum Market Model addresses these limitations through a systematic, quantitative approach that integrates 16 verified fundamental indicators across five critical market dimensions. The model builds upon established financial engineering principles while incorporating petroleum-specific market dynamics and adaptive learning mechanisms.
## 2. Theoretical Framework
### 2.1 Market Efficiency and Information Integration
The model operates under the assumption of semi-strong market efficiency, where fundamental information is gradually incorporated into prices with varying degrees of lag (Fama, 1970). The petroleum market's unique characteristics, including storage costs, transportation constraints, and geopolitical risk premiums, create opportunities for fundamental analysis to provide predictive value (Kilian, 2009).
### 2.2 Multi-Factor Asset Pricing Theory
Drawing from Ross's (1976) Arbitrage Pricing Theory, the model treats petroleum prices as driven by multiple systematic risk factors. The five-factor decomposition (Supply, Inventory, Demand, Trade, Sentiment) represents economically meaningful sources of systematic risk in petroleum markets (Chen et al., 1986).
## 3. Methodology
### 3.1 Data Sources and Quality Framework
The model integrates 16 fundamental indicators sourced from verified TradingView economic data feeds:
Supply Indicators:
- US Oil Production (ECONOMICS:USCOP)
- US Oil Rigs Count (ECONOMICS:USCOR)
- API Crude Runs (ECONOMICS:USACR)
Inventory Indicators:
- US Crude Stock Changes (ECONOMICS:USCOSC)
- Cushing Stocks (ECONOMICS:USCCOS)
- API Crude Stocks (ECONOMICS:USCSC)
- API Gasoline Stocks (ECONOMICS:USGS)
- API Distillate Stocks (ECONOMICS:USDS)
Demand Indicators:
- Refinery Crude Runs (ECONOMICS:USRCR)
- Gasoline Production (ECONOMICS:USGPRO)
- Distillate Production (ECONOMICS:USDFP)
- Industrial Production Index (FRED:INDPRO)
Trade Indicators:
- US Crude Imports (ECONOMICS:USCOI)
- US Oil Exports (ECONOMICS:USOE)
- API Crude Imports (ECONOMICS:USCI)
- Dollar Index (TVC:DXY)
Sentiment Indicators:
- Oil Volatility Index (CBOE:OVX)
### 3.2 Data Quality Monitoring System
Following best practices in quantitative finance (Lopez de Prado, 2018), the model implements comprehensive data quality monitoring:
Data Quality Score = Σ(Individual Indicator Validity) / Total Indicators
Where validity is determined by:
- Non-null data availability
- Positive value validation
- Temporal consistency checks
### 3.3 Statistical Normalization Framework
#### 3.3.1 Z-Score Normalization
The model employs robust Z-score normalization as established by Sharpe (1994) for cross-indicator comparability:
Z_i,t = (X_i,t - μ_i) / σ_i
Where:
- X_i,t = Raw value of indicator i at time t
- μ_i = Sample mean of indicator i
- σ_i = Sample standard deviation of indicator i
Z-scores are capped at ±3 to mitigate outlier influence (Tukey, 1977).
#### 3.3.2 Percentile Rank Transformation
For intuitive interpretation, Z-scores are converted to percentile ranks following the methodology of Conover (1999):
Percentile_Rank = (Number of values < current_value) / Total_observations × 100
### 3.4 Exponential Smoothing Framework
Signal smoothing employs exponential weighted moving averages (Brown, 1963) with adaptive alpha parameter:
S_t = α × X_t + (1-α) × S_{t-1}
Where α = 2/(N+1) and N represents the smoothing period.
### 3.5 Dynamic Threshold Optimization
The model implements adaptive thresholds using Bollinger Band methodology (Bollinger, 1992):
Dynamic_Threshold = μ ± (k × σ)
Where k is the threshold multiplier adjusted for market volatility regime.
### 3.6 Composite Score Calculation
The fundamental score integrates component scores through weighted averaging:
Fundamental_Score = Σ(w_i × Score_i × Quality_i)
Where:
- w_i = Normalized component weight
- Score_i = Component fundamental score
- Quality_i = Data quality adjustment factor
## 4. Implementation Architecture
### 4.1 Adaptive Parameter Framework
The model incorporates regime-specific adjustments based on market volatility:
Volatility_Regime = σ_price / μ_price × 100
High volatility regimes (>25%) trigger enhanced weighting for inventory and sentiment components, reflecting increased market sensitivity to supply disruptions and psychological factors.
### 4.2 Data Synchronization Protocol
Given varying publication frequencies (daily, weekly, monthly), the model employs forward-fill synchronization to maintain temporal alignment across all indicators.
### 4.3 Quality-Adjusted Scoring
Component scores are adjusted for data quality to prevent degraded inputs from contaminating the composite signal:
Adjusted_Score = Raw_Score × Quality_Factor + 50 × (1 - Quality_Factor)
This formulation ensures that poor-quality data reverts toward neutral (50) rather than contributing noise.
## 5. Usage Guidelines and Best Practices
### 5.1 Configuration Recommendations
For Short-term Analysis (1-4 weeks):
- Lookback Period: 26 weeks
- Smoothing Length: 3-5 periods
- Confidence Period: 13 weeks
- Increase inventory and sentiment weights
For Medium-term Analysis (1-3 months):
- Lookback Period: 52 weeks
- Smoothing Length: 5-8 periods
- Confidence Period: 26 weeks
- Balanced component weights
For Long-term Analysis (3+ months):
- Lookback Period: 104 weeks
- Smoothing Length: 8-12 periods
- Confidence Period: 52 weeks
- Increase supply and demand weights
### 5.2 Signal Interpretation Framework
Bullish Signals (Score > 70):
- Fundamental conditions favor price appreciation
- Consider long positions or reduced short exposure
- Monitor for trend confirmation across multiple timeframes
Bearish Signals (Score < 30):
- Fundamental conditions suggest price weakness
- Consider short positions or reduced long exposure
- Evaluate downside protection strategies
Neutral Range (30-70):
- Mixed fundamental environment
- Favor range-bound or volatility strategies
- Wait for clearer directional signals
### 5.3 Risk Management Considerations
1. Data Quality Monitoring: Continuously monitor the data quality dashboard. Scores below 75% warrant increased caution.
2. Regime Awareness: Adjust position sizing based on volatility regime indicators. High volatility periods require reduced exposure.
3. Correlation Analysis: Monitor correlation with crude oil prices to validate model effectiveness.
4. Fundamental-Technical Divergence: Pay attention when fundamental signals diverge from technical indicators, as this may signal regime changes.
### 5.4 Alert System Optimization
Configure alerts conservatively to avoid false signals:
- Set alert threshold at 75+ for high-confidence signals
- Enable data quality warnings to maintain system integrity
- Use trend reversal alerts for early regime change detection
## 6. Model Validation and Performance Metrics
### 6.1 Statistical Validation
The model's statistical robustness is ensured through:
- Out-of-sample testing protocols
- Rolling window validation
- Bootstrap confidence intervals
- Regime-specific performance analysis
### 6.2 Economic Validation
Fundamental accuracy is validated against:
- Energy Information Administration (EIA) official reports
- International Energy Agency (IEA) market assessments
- Commercial inventory data verification
## 7. Limitations and Considerations
### 7.1 Model Limitations
1. Data Dependency: Model performance is contingent on data availability and quality from external sources.
2. US Market Focus: Primary data sources are US-centric, potentially limiting global applicability.
3. Lag Effects: Some fundamental indicators exhibit publication lags that may delay signal generation.
4. Regime Shifts: Structural market changes may require model recalibration.
### 7.2 Market Environment Considerations
The model is optimized for normal market conditions. During extreme events (e.g., geopolitical crises, pandemics), additional qualitative factors should be considered alongside quantitative signals.
## References
Baumeister, C., & Kilian, L. (2016). Forty years of oil price fluctuations: Why the price of oil may still surprise us. *Journal of Economic Perspectives*, 30(1), 139-160.
Bollinger, J. (1992). *Bollinger on Bollinger Bands*. McGraw-Hill.
Brown, R. G. (1963). *Smoothing, Forecasting and Prediction of Discrete Time Series*. Prentice-Hall.
Chen, N. F., Roll, R., & Ross, S. A. (1986). Economic forces and the stock market. *Journal of Business*, 59(3), 383-403.
Conover, W. J. (1999). *Practical Nonparametric Statistics* (3rd ed.). John Wiley & Sons.
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. *Journal of Finance*, 25(2), 383-417.
Hamilton, J. D. (2009). Understanding crude oil prices. *Energy Journal*, 30(2), 179-206.
Kilian, L. (2009). Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market. *American Economic Review*, 99(3), 1053-1069.
Lopez de Prado, M. (2018). *Advances in Financial Machine Learning*. John Wiley & Sons.
Ross, S. A. (1976). The arbitrage theory of capital asset pricing. *Journal of Economic Theory*, 13(3), 341-360.
Sharpe, W. F. (1994). The Sharpe ratio. *Journal of Portfolio Management*, 21(1), 49-58.
Tukey, J. W. (1977). *Exploratory Data Analysis*. Addison-Wesley.
開源腳本
本著TradingView的真正精神,此腳本的創建者將其開源,以便交易者可以查看和驗證其功能。向作者致敬!雖然您可以免費使用它,但請記住,重新發佈程式碼必須遵守我們的網站規則。
免責聲明
這些資訊和出版物並不意味著也不構成TradingView提供或認可的金融、投資、交易或其他類型的意見或建議。請在使用條款閱讀更多資訊。
開源腳本
本著TradingView的真正精神,此腳本的創建者將其開源,以便交易者可以查看和驗證其功能。向作者致敬!雖然您可以免費使用它,但請記住,重新發佈程式碼必須遵守我們的網站規則。
免責聲明
這些資訊和出版物並不意味著也不構成TradingView提供或認可的金融、投資、交易或其他類型的意見或建議。請在使用條款閱讀更多資訊。