Análise comparativa de estratégias de day trading: ORB isolado vs. pair trading com seleção por grafos

Abstract

The search for investment strategies offering consistent returns uncorrelated with market indices is a central challenge in quantitative finance. This work proposes and evaluates a hybrid methodology for Day Trading operations in the Brazilian stock market, combining the volatility capture of the Opening Range Breakout (ORB) strategy with the directional risk mitigation of Pair Trading. The methodological differentiator lies in the development of a pair selection algorithm based on Graph Theory, which replaces traditional exhaustive search with an adaptive spreading heuristic, weighted by fundamental similarity and controlled by the correlation of the Higuchi Fractal Dimension (HFD). Empirical validation was conducted through rigorous backtesting on a universe of 83 assets from B3, covering the period from 2018 to 2024. The results demonstrated that the hybrid strategy was able to generate a Cumulative Return of 301.72%, significantly outperforming the CDI and Ibovespa, with a Win Rate exceeding 51% and a positive Payoff Ratio. However, risk analysis revealed high annualized volatility, evidencing that while the fractal-based heuristic approach is effective in generating Alpha, the model’s stability across different market regimes requires the future incorporation of secondary artificial intelligence filters.