Centering price data improves predicting stock returns from intraday bars
Centering Drives Normalization Gains: Price-Offset Nuisances in Cross-Sectional Return Prediction
Computational Engineering, Finance, and Science
Summary
Predicting how stock prices will move during the day is tricky because the price level of each stock can confuse the models. The authors found that simply removing the average price from the data, called centering, makes prediction models work better. Changing how much the data is scaled or using different model types did not help as much. This means that focusing on removing price offsets is key for better stock return predictions using detailed price data.
What this means in practice
- •For quantitative trading teams: Improve cross-sectional stock return predictions by centering price data in models using high-frequency intraday price inputs.
- •For financial data engineers: Design data preprocessing pipelines that remove price offsets rather than only scaling to boost predictive quality of intraday price features.
Tested on one dataset.
Authors
Mingju Chen, Qianhui Liu, Yui Lo, Yuanhang Liu
Abstract
Cross-sectional return prediction from raw intraday bars is sensitive to each instrument price level, an additive nuisance under a return-ranking hypothesis. We test whether removing this offset, rather than rescaling amplitudes or changing the encoder, explains gains on a point-in-time CSI 300 five-minute panel. Eight parameter-matched encoders are evaluated with and without RevIN normalization; a parameter-free ladder then separates identity, scale-only, centering, last-value referencing, differencing, and standardization across all fields and restricted channels. Centering drives the reliable effect, while scale-only normalization does not help. All eight paired effects are positive and survive Holm correction on raw rank IC, after style residualization, and after additionally residualizing on short-term reversal. Among six stronger encoders, normalized IC is 0.0830-0.0939 and gains are 0.0376-0.0567. Price-only standardization retains 93-101% of the all-field gain. These results place the main effect in transformed price-channel offset removal rather than amplitude scaling or encoder choice.