Please rotate your device to landscape mode to view the charts.

Direction-of-change forecasting in commodity futures markets

Journal: International Review of Financial Analysis

Date: 2021

Author: Jiadong Liu, Fotis Papailias, Barry Quinn

Abstract:
This paper examines direction-of-change predictability in commodity futures markets using a variety of binary probabilistic techniques. As well as traditional techniques, we apply Variable Length Markov Chain (VLMC) analysis, an innovative technique popularised in computational biology when predicting DNA sequences (Buhlmann & Wyner, 1999). To the best of our knowledge, this is the first application of VLMC in finance. Our results show that both VLMC and technical analysis methods provide strong predictability of the direction-of-change of commodity returns, with annualised mean returns of approximately 8%, substantially higher than the passive long strategy. Our results suggest that a short-term learning effect is present in commodities market which can be exploited using innovative direction-of-change forecasting techniques.

Link: Google Scholar

Background and Context

Research Motivation

Traditional financial forecasting focuses on predicting exact returns, but investors are more interested in whether they should buy or sell assets based on predicted market direction.

Study Focus

This research examines direction-of-change predictability in commodity futures markets using innovative binary probabilistic techniques including Variable Length Markov Chain analysis.

Methodology

The study analyzes weekly returns of 24 commodity futures indices from 1971-2018 using Return Signal Momentum, dynamic probit models, and Variable Length Markov Chain approaches.

Superior Performance of RSM and VLMC Models Using 52-Week Training Window

  • Shows the success rates of different forecasting models using 52-week training window
  • VLMC-BS (bootstrapped) and RSM models achieve highest success rates above 52%
  • Traditional probit models show lower predictive accuracy around 51%

Declining Model Performance with Longer Training Windows

  • Shows how model performance changes with different training window lengths
  • Performance peaks at 52-week window and gradually declines with longer windows
  • Suggests short-term market patterns are more predictable than long-term trends

Superior Investment Returns from RSM and VLMC Strategies

  • Compares annualized returns of different trading strategies using 52-week window
  • RSM and VLMC-BS strategies achieve highest returns of ~7.8%
  • All strategies outperform passive buy-and-hold approach

Sector-Specific Predictability in Commodity Markets

  • Shows prediction success rates across different commodity sectors
  • Metals and energy sectors show higher predictability
  • Agricultural commodities show relatively lower predictability

Cumulative Strategy Performance Over Time

  • Shows total investment returns over the full study period
  • RSM and VLMC-BS strategies achieve ~30x cumulative returns
  • Demonstrates significant long-term outperformance versus traditional approaches

Contribution and Implications

  • First application of Variable Length Markov Chain (VLMC) analysis in financial markets shows promising results for predicting commodity price movements
  • Demonstrates that short-term (52-week) market patterns are more predictable than long-term trends in commodity markets
  • Provides evidence that technical trading strategies based on direction-of-change forecasting can generate significant excess returns

Data Sources

  • Success rate comparison chart based on Table 3
  • Training window performance chart based on Table 3
  • Investment returns chart based on Table 5
  • Sector-specific success rates derived from Table 4
  • Cumulative performance chart based on Table 5