Umeå universitets logga

Technical analysis remains widely used in financial markets, despite the continuing debate about whether chart patterns contain genuine predictive information or mainly reflect broader market exposure, data-snooping effects, and favorable backtesting assumptions. This thesis investigates whether commonly used technical chart-pattern strategies generate economically meaningful and risk-adjusted abnormal returns when evaluated under systematic and reproducible conditions.

The study uses approximately 20 years of daily data for 100 stocks from the S\&P 500. Historical OHLC price data are cleaned, validated, and corrected for basic consistency errors before being used in a rule-based backtesting framework. Two main strategy families are implemented: moving-average crossover strategies and support-and-resistance breakout strategies based on confirmed pivot points. Each strategy is tested across multiple parameter combinations and evaluated in both long and short form. The backtests include explicit transaction-cost assumptions, including bid-ask spread, slippage, and broker fees. In addition, two portfolio specifications are compared: one where inactive capital earns the daily risk-free rate and one where inactive capital receives no return.

The empirical performance is assessed using CAPM alpha and beta, allowing the strategy returns to be separated into abnormal performance and systematic market exposure. The results show that the moving-average strategies perform consistently poorly, with negative CAPM alphas across most specifications. Short strategies also perform poorly, producing negative risk-adjusted returns and negative total trade-level outcomes. Long support-and-resistance breakout strategies generate the most favorable results, particularly under the specification where inactive capital earns the risk-free rate. However, these positive results weaken substantially when the bond allocation is removed.

Overall, the findings do not provide strong evidence that the tested chart-pattern strategies generate robust abnormal returns after accounting for transaction costs and market exposure. The results suggest that apparent profitability in technical trading rules is sensitive to implementation assumptions, portfolio construction, and risk adjustment.