Abstract Although the extant empirical strands studying herding differ from divergent objects to periods, they are common in a higher probability of investment behavior abnormality. There is scant empirical evidence on herding across cross-section of stocks in a market over a general period of business cycle. This paper tests herd behavior in a transparent and order-driven market. We propose to the literature (1) a modification of the most commonly applied herding measure, (2) that investors tend to herd more based on fundamental analysis (past P/E ratio) than on technical analysis (past returns), (3) that, according to the prospect theory, the reward structure of investors is asymmetric based on the discrepancy between the sell-side herding model and buy-side herding model and (4) that, informational cascade can be identified through the combined study of probability of information trading (PIN) and herding measure. In general, our analyses agree with former literature in that herding is more prevalent in small stocks and in selling stocks. Most importantly, our results verify the existence of the well-known phenomenon of informational cascade in our sample, which highlight a crucial role of “fashion leaders”, especially when more informed investors trade with noise.