Behavioral risks including investor sentiment plays an important roles in Asia’s stock market behaviors but theoretically and empirically not well understood in traditional conditional means statistic partly due to its heterogeneity roles. In contrast to the existing studies, this paper examine the heterogeneity roles of investor sentiment proxies (i.e. CSI, BCS, and FKLI) in influencing various aggregate stock market indices in Malaysia. The proposed sentiment proxies relation to stock returns are statistically analyzed using mean-based (OLS) and quantile based (QR) regression methods to uncover the relationships accross full range of stock market returns conditional distributions. In the analysis, we examine the possible multidimensional association of various conditional of sentiments (i.e. average, sentiment states, market states, and interaction between sentiment and market states) on size (i.e. big firms vs. small firms) and industry (i.e. defensive vs. speculative) segmented data. The OLS analysis does not provides conclusive significant association of sentiment to returns but the QR analysis reveal emerging patterns of sentiment heterogeneity roles. Specifically, the QR analysis reveal an asymmetric association between sentiment to returns with U-curve pattern from negative magnitude in the lower quantiles and positive magnitude in the upper quantiles. The findings not only consistent with the current hypothesis that sentiment risk is strongly affected the small firms and speculative industry but also the big firms. The sentiment-return associations are statistically significant in extreme lower quantiles and in extreme upper quantiles of return distributions. These patterns are consistent with theoretical postulates of prospect theory and evidence of assymetric overreaction of Asia investors to sentiment news. Overall, the findings from this paper provides new insights for theoretical understanding and practical application of sentiment risk in stock market.