A successful prediction error expansion (PEE) based reversible data hiding (RDH) algorithm requires an useful pixel prediction algorithm. You can find a plethora of pixel prediction methods in books and online. Gradients are the key issue for predicting the current pixel. Nowadays researchers are more focused on gradients for better predicting the current pixel. The gradient can be used for better analyzing the pixel information. In this study, a novel method for improving current pixel prediction is presented employing shades in the image and rhombus context on \(5 \times 5\) neighborhood. Based on the local complexity (LoCo) of the pixel, An innovative AHBS has been utilized to incorporate additional data while minimizing distortion. Information retrieval process has been experimented using gray scale images. Findings from the experiment show that the proposed approach is superior than other existing approaches. This method can be applied to various business applications where information hiding plays a crucial role.