Signal Fusion and Empirical Research on Volatility Factors in High-Frequency Trading of Crypto Assets

Due to the wide application of high-frequency trading in the crypto asset market, and because volatility factors can accurately reflect the characteristics of price changes, they have long attracted attention from both research and practice. However, a single volatility factor is often disturbed by market noise and has a small adjustment range. In this paper, multiple high-frequency volatility factors are constructed, such as historical volatility, realized volatility, and jump volatility, and three fusion techniques are designed, namely linear weighting, statistical dimension reduction, and machine learning fusion methods. Through empirical tests using the transit-by-transaction data of BTC and ETH, the results show that the comprehensive signal strategy outperforms the single-factor strategy in terms of prediction effect, stability of positive returns, and risk control, demonstrating obvious trading advantages.