In a new report sent to Rigzone this week, Standard Chartered revealed that it is launching a machine learning model for short-term Brent price forecasting.
Called SCORPIO (Standard Chartered Oil Research Price Indicator), Standard Chartered describes the new launch as a proprietary tree-based model designed to generate a forecast of Brent crude oil spot prices over a one-week period.
“The machine learning model programmatically collects and analyzes a set of available data and features, weighing the information into a meaningful signal,” Standard Chartered said in the report.
“It allows for an element of explainability and helps decouple market sentiment from fundamentals,” he added.
The model incorporates features including high-frequency data points, price information for crude and refined products, technical indicators, positioning data, global inventories, implied demand, imports and exports, as well as non-oil data such as the USD strength, PMI and other macroeconomic inputs, Standard Chartered described in the report.
“The model also allows us to separate the effects of unexpected events and macro market sentiment from oil market fundamentals to explain short-term price movements,” the company said in the report.
The latest iteration of SCORPIO shows a “statistically significant” directional accuracy of 67.3 percent, Standard Chartered noted in the report.
“The mean absolute error is less than one standard deviation of the observations over the last 52 weeks, and the standard deviation of the error is also less than the standard deviation of the observations,” the company added.
Construction of the model
Explaining the construction of the model in the report, Standard Chartered said it first looked at the most basic set of features: “high-frequency data points used by economists to model the fundamentals of supply and demand, as well like the technical characteristics of the market”.
“We then added more data sets, including non-oil-specific macroeconomic data. We tested several alternative data sets. Some showed considerable significance, while others showed little or no explanatory or predictive power for oil prices in the short term,” they added.
“A prominent example was news sentiment data, where news content on chosen topics or keywords is collected and processed to determine whether overall sentiment is negative, neutral, or positive,” they continued.
“While programmatic analysis of news reports can provide an advantage in high-frequency trading (for example, the model could react immediately to news such as a drone attack on a production facility), this did not provide any additional value to the model on a daily to weekly basis,” they stated.
The company said in the report that the model’s overall performance was measured on back-tested predictive performance over a period before.
“Performance is measured using value-based metrics (mean absolute error) and directional metrics (up, near zero or down), where ranking and evaluation metrics were used,” Standard Chartered said in the report.
“We draw error bands around all predictions based on quantile regression methods (with the same characteristics). A detailed report of characteristics can be generated that explains the derivation of predicted price movements, grouped by thematic categories,” he add.
“Due to data availability, the time frame for the data sources used is from 2018. These dates also ensure that pre-Covid, Covid and post-Covid business environments are captured, in addition to the Russian invasion of Ukraine in 2022,” he continued.
Standard Chartered said in the report that it will continue to refine the model over time, “adding new features if they are shown to improve performance.”
limitations
Because not all known drivers are captured in reliable data pipelines, a machine learning model can be considered a simplified representation of a subset of known drivers, Standard Chartered said in the report.
“However, it is still vulnerable to so-called ‘black swan’ events – unpredictable events that would not be detected within our set of indicators and could significantly affect short-term price movements,” the company added.
“In the oil space, these events could include rapidly developing severe hurricanes, geopolitical developments or acts of terrorism, producer policy decisions or broader macroeconomic events such as bank collapses,” he continued.
“From early March to mid-April 2023, there were two significant ‘black swan’ events that SCORPIO failed to forecast. However, it was able to accurately forecast the price direction of other indicators in the weeks following these events,” the company said.
Black swan events highlighted in the report were the collapse of Silicon Valley Bank and the implementation of additional voluntary production cuts by some OPEC+ members earlier this year.
SCORPIO Forecast
In a separate report sent to Rigzone this week, Standard Chartered revealed that SCORPIO expects a weekly price increase of $2.1 per barrel for first-month Brent to settlement on October 2.
“The upward forecast would have been greater if it were not for the speculative positioning; the model is interpreting the strength of the Standard Chartered Money Managers Positioning Index as a pivot point indicator,” Standard Chartered noted in this report.
“SCORPIO also sees USD strength weighing on the move higher, with the index jumping from the 87th percentile over the past five years to the 91st percentile,” the company added.
In this report, Standard Chartered highlighted that one of the potential uses of SCORPIO is to indicate whether speculative positioning is so overextended as to become a dominant price driver.
“For the week ahead, SCORPIO sees the positioning and strength of the USD as frictions, but not yet enough to lower prices,” the company said in the report.
Market opinion
Prior to the launch of Standard Chartered’s SCORPIO model, Rigzone polled a number of market participants if AI can predict the price of oil.
The answer to that question was no, according to Alex Stevens, the Policy and Communications Manager at the Institute for Energy Research (IER), which describes itself as a non-profit organization that conducts intensive research and analysis on the functions , operations, and government regulation of global energy markets.
Responding to the same question, Hussein Shel, Director, Chief Technologist and Head of Upstream Power and Utilities at Amazon Web Services (AWS), said: “Machine learning and intel technologies artificial intelligence, including generative AI and similar language models, are not specifically designed to predict financial markets, including oil prices.”
When Al Salazar, Senior Vice President of Enverus Intelligence Research (EIR) was asked if AI can predict the price of oil, the EIR representative told Rigzone that “AI could have some advantages in terms of of data and computing power that conventional forecasters don’t have.”, but added that “one thing AI could struggle with is properly timing OPEC actions alongside geopolitically driven supply disruptions.”
To contact the author, please send an email andreas.exarcheas@rigzone.com
