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

Charting by Machines

Journal: Journal of Financial Economics

Date: 20240301

Author: Murray, Scott; Xia, Yusen; Xiao, Houping

Abstract:
We test the efficient market hypothesis by using machine learning to forecast stock returns from historical performance. These forecasts strongly predict the cross-section of future stock returns. The predictive power holds in most subperiods and is strong among the largest 500 stocks. The forecasting function has important nonlinearities and interactions, is remarkably stable through time, and captures effects distinct from momentum, reversal, and extant technical signals. These findings question the efficient market hypothesis and indicate that technical analysis and charting have merit. We also demonstrate that machine learning models that perform well in optimization continue to perform well out-of-sample.

Link: Google Scholar

Key Findings

Strong Cross-Sectional Return Predictability

Machine learning forecasts strongly predict future stock returns with an average excess return of 1.08% per month for the long-short portfolio, demonstrating significant violations of market efficiency

Stable Through Time & Large Stocks

Predictive power remains remarkably stable across different time periods and among the largest 500 stocks, with the long-short portfolio generating 0.72% monthly returns for top 500 stocks

Complex Nonlinear Patterns

The ML forecasting function captures substantial nonlinear and interaction effects distinct from traditional momentum and reversal patterns

Portfolio Performance Across Time

  • Strong performance across most subperiods from 1963-2022
  • Average monthly returns range from 1.15% to 1.71% in most periods
  • Only period of weak performance was 2005-2014 (-0.08%)

Performance Across Market Cap Groups

  • Strategy remains effective even among the largest stocks
  • Returns decrease only modestly as market cap increases
  • Top 500 stocks still generate significant 0.72% monthly returns

Factor Model Performance

  • Strategy generates significant alpha across different factor models
  • Monthly alphas range from 0.45% to 1.24% depending on the model
  • Results robust to controlling for standard risk factors

Contribution and Implications

  • Provides strong evidence against the weak-form market efficiency hypothesis by showing profitable trading strategies based solely on past price patterns
  • Demonstrates that technical analysis and charting have more merit than previously acknowledged in academic literature
  • Shows machine learning can effectively identify complex nonlinear patterns in stock returns that are distinct from known factors

Data Sources

  • Subperiod performance chart based on Table 5 showing average returns across different time periods
  • Market cap analysis chart based on Table 6 showing returns across different size groups
  • Factor model performance chart based on Table 4 Panel A showing alphas across different factor models