TL;DR:
- Standard Chartered introduces SCORPIO, a machine learning model for predicting Brent crude oil prices in a one-week timeframe.
- SCORPIO leverages diverse data sources, including high-frequency data, pricing information, technical indicators, and macroeconomic inputs.
- The model boasts a 67.3% directional accuracy and lower mean absolute error compared to historical observations.
- SCORPIO’s construction involved a thorough examination of various data sets, with a focus on fundamental and macroeconomic factors.
- The model acknowledges its limitations, particularly in predicting ‘black swan’ events.
- SCORPIO’s latest forecast indicates a week-on-week price increase of $2.1 per barrel for front-month Brent, attributing it to speculative positioning and USD strength.
- Expert opinions vary on AI’s ability to predict oil prices, citing challenges in timing OPEC actions and geopolitical supply disruptions.
Main AI News:
In an exclusive report shared with Rigzone this week, Standard Chartered makes a significant stride in the world of energy finance by introducing SCORPIO (Standard Chartered Oil Research Price Indicator). This proprietary tree-based machine learning model is poised to revolutionize near-term Brent crude spot price forecasting, offering a valuable tool for investors and market analysts.
SCORPIO stands out as a sophisticated model meticulously designed to provide one-week forecasts for Brent crude spot prices. By programmatically collecting and analyzing a wide array of available data and features, it leverages the power of machine learning to distill this information into meaningful signals. One of its key advantages lies in its ability to decouple market sentiment from underlying fundamentals, enhancing the clarity of decision-making.
This advanced model incorporates a comprehensive range of features, including high-frequency data points, pricing data for crude and refined products, technical indicators, positioning data, global stocks, implied demand, imports, exports, and non-oil data such as USD strength, PMIs, and other macroeconomic inputs. By doing so, Standard Chartered aims to provide a holistic view of the complex factors influencing Brent crude prices.
Furthermore, SCORPIO’s versatility extends to isolating the impact of unforeseen events and macroeconomic market sentiment, enabling a clearer understanding of short-term price fluctuations. With a statistically significant directional accuracy of 67.3 percent, SCORPIO is already proving its mettle. Its mean absolute error, when compared to the observations of the last 52 weeks, is lower, and the error standard deviation is also within bounds.
Constructing this groundbreaking model involved a meticulous process. It began with the examination of fundamental features such as high-frequency data points, commonly used by economists to model supply and demand dynamics, along with technical market features. To enhance its predictive power, Standard Chartered incorporated non-oil-specific macroeconomic data and thoroughly explored various alternative data sets. Notably, news sentiment data, although valuable in high-frequency trading scenarios, did not significantly contribute to the model’s daily to weekly predictions.
Standard Chartered’s commitment to transparency is evident in its approach to measuring model performance. Performance assessments include both value-based metrics, such as mean absolute error, and directional metrics, classifying price movements as up, near zero, or down. Additionally, error bands are meticulously drawn around predictions based on quantile regression methods, providing a detailed feature report that explains the predicted price movements.
It is noteworthy that the model’s data sources extend from 2018 onwards, capturing significant market events, including the pre-Covid, Covid, and post-Covid trading landscapes, as well as Russia’s invasion of Ukraine in 2022.
Standard Chartered affirms its dedication to enhancing the model’s performance over time by incorporating new features that demonstrate improved predictive capabilities.
However, SCORPIO does have its limitations. Not all known drivers can be reliably captured in data pipelines, leaving the model vulnerable to unforeseeable ‘black swan’ events—events that lie outside the scope of its indicators yet can significantly impact short-term price movements. These events might include severe hurricanes, geopolitical developments, producer policy decisions, or broader macroeconomic events.
In early 2023, SCORPIO encountered two such ‘black swan’ events: the collapse of Silicon Valley Bank and the implementation of additional voluntary output cuts by select OPEC+ members. While these events were unforeseen, SCORPIO efficiently adapted by accurately forecasting subsequent price movements based on other indicators.
SCORPIO’s most recent forecast, shared in a separate report to Rigzone, predicts a week-on-week price increase of $2.1 per barrel for front-month Brent, culminating in settlement on October 2. This upward trajectory could have been more substantial were it not for speculative positioning. SCORPIO interprets the Standard Chartered money-manager positioning index’s strength as a pivot point indicator. Additionally, it identifies USD strength as a factor influencing the price increase.
One intriguing aspect of SCORPIO is its potential to assess whether speculative positioning becomes a dominant price factor. While it recognizes positioning and USD strength as potential drags for the upcoming week, it does not anticipate a significant downward pull on prices.
In seeking expert opinions on AI’s ability to predict oil prices, we reached out to key figures in the industry. While AI and machine learning technologies offer advantages in terms of data processing and computing power, experts, including Alex Stevens from the Institute for Energy Research (IER), Hussein Shel from Amazon Web Services (AWS), and Al Salazar from Enverus Intelligence Research (EIR), agree that AI struggles to precisely time OPEC actions and geopolitical supply outages.
Conclusion:
Standard Chartered’s SCORPIO represents a significant advancement in the energy finance sector, offering a powerful tool for short-term Brent crude oil price forecasting. Its high accuracy and comprehensive data integration provide valuable insights for market participants. However, the model’s limitations underscore the ongoing need for human judgment in navigating unforeseen ‘black swan’ events, ensuring that AI augments rather than replaces human expertise in the complex world of energy finance.
