建筑装饰指数趋势跟踪模型效果点评

• The model is named "Building Decoration Index Trend Tracking Model" and is designed based on the assumption that the price trend of the target has strong local continuity, with reversal periods being significantly shorter than trend continuation periods. It assumes that narrow-range consolidation will continue the previous trend. When a major trend is present, a short observation window will reflect the local trend, and reversals will show price changes exceeding the range of random fluctuations, thus filtering out random noise[2][3] - The model targets the SW First-Level Building Decoration Index, with raw data retained for preprocessing. It operates on a long-short signal dimension[3] - The specific algorithm involves calculating the difference between the closing price on day T and day T-20 ($del$), as well as the volatility ($Vol$) from day T-20 to day T (excluding T). If the absolute value of $del$ exceeds $N$ times $Vol$, it indicates a trend breakout, with the trend direction determined by the sign of $del$. Otherwise, the trend direction follows that of day T-1. For this model, $N$ is set to 1 to capture smaller opportunities in the more volatile stock market. The model evaluates combined long-short returns as the final performance metric[3] - The model's backtesting period spans from March 7, 2023, to March 18, 2025[3] - The model's performance evaluation indicates that it achieved its highest net value during the period from March 7, 2023, to January 22, 2024. However, from January 22, 2024, to September 26, 2024, the net value declined due to market conditions. Subsequently, the net value returned to near its historical high but entered a period of oscillation. The model demonstrates relatively low annualized returns and prolonged drawdowns in the later stages, making it unsuitable for direct application to the SW First-Level Building Decoration Index[4] - The model's backtesting results include the following metrics: annualized return of 4.39%, annualized volatility of 23.96%, Sharpe ratio of 0.18, maximum drawdown of 22.47%, and total return of -12.25% during the evaluation period[3]