Centering price data improves stock return predictions over scaling
Centering Drives Normalization Gains: Price-Offset Nuisances in Cross-Sectional Return Prediction
Computational Engineering, Finance, and Science
Summary
Predicting stock returns using detailed price data can be tricky because the absolute price level of each stock adds unwanted noise. The authors found that simply removing this constant price offset, called centering, helps improve prediction accuracy much more than adjusting the scale of the prices or changing the prediction method. They tested several prediction models and showed that centering consistently improved results across different conditions. This suggests that focusing on correcting price offsets is more important than other common data adjustments for return prediction.
cross-sectional return predictionintraday barsprice offsetcenteringnormalizationencoderrank ICstyle residualizationshort-term reversalreturn-ranking hypothesis
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.