Algorithms comparison on intraday index return prediction：evidence from China

ABSTRACT

We introduce the fading memory recursive least squares (FM-RLS) and rolling window ordinary least squares (RW-OLS) methods to predict CSI 300 intraday index return in Chinese stock market. Empirical results show that the performances are better than that of same sign method. The additional profit is mainly from two conflict signals, with one amplitude far greater than the other.

KEYWORDS:

• Least squares model
• rolling window ordinary least squares model
• fading memory recursive least squares model
• same sign model
• predictability

JEL CLASSIFICATION:

• C22
• C52
• C53

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [11631013, 11971372, and 11801433].