Data and code for "Can artificial intelligence improve technical-indicator trading strategies? Evidence from Taiwan index futures"

This record provides data and code supporting the study “Can artificial intelligence improve technical-indicator trading strategies? Evidence from Taiwan index futures.” The materials cover 91 Taiwan index futures strategies across four architectures: Traditional Rules, AI Single-Indicator, Hybrid, and AI Multi-Indicator. Formal evaluation covers 1,948 trading observations from January 2, 2018 to December 31, 2025; supplementary out-of-sample evaluation covers 116 observations from January 2 to June 30, 2026. The archive includes source code, environment requirements, model probabilities and gates, strategy positions, transaction costs, returns, equity series, performance and ranking tables, statistical results, figures, and verification instructions. Raw TAIFEX market data and price-based feature matrices are not included. Data-source information, acquisition code, and source-file checksums are provided for users obtaining the original data separately. Full retraining requires those market inputs. Original research outputs are licensed under CC BY 4.0, and original software is licensed under the MIT License. Third-party components retain their respective licenses. See LICENSING.md within the archive for details.