Behavioral Aspects of Market Anomalies Anomaly is defined as ‘something that deviates from what is standard, normal, or expected’ by Oxford dictionary (2016). George and Elton (2001) has defined market anomaly as a new or unexpected phenomenon in relation to any theory, model or hypothesis. The founder of behavioural finance, Tversky and Kahneman (1986), suggested that the market anomalies are the indicators of inefficient markets, which might either occur only once and disappear, or occur regularly. The school of Neoclassical Finance is currently the prevailing paradigm in the finance field. In neo-classical finance, the investors make statistical judgement based on Von Neumann–Morgenstern utility theorem and Bayesian techniques (Ackert and
…show more content…
The ‘loyalists’ support Efficient Market Hypothesis (EMH) by pointing out the problem of data mining or market imperfections, the ‘revisionists’ support EMH with time-varying risk premiums, whereas the ‘heretics’ challenge EMH and believe that the market is irrational and the psychological factors could influence the pricing of securities, where the abnormal returns can be earned by using specific trading strategies. The rationalists and revisionists believe that the stock returns cannot be predicted and thus the superior performance of specific investment strategies, for example the momentum and contrarian trading, and value investing are not an anomaly. They suggest that the superior returns are a result by chance instead (Black 1993). Besides, they argue that the abnormal returns are the result of higher risk premium (Fama and French, 1996). For instance, the value stocks are risker than the growth stocks, and thus the higher risk from investing in value stocks has generated the superior profits. However, a significant abnormal return can still be found even after adjusting for risk (Chopra, Lakonishok and Ritter …show more content…
Data mining occurs when a researcher tries many approaches to do a particular study, such as different time periods, different models and different combinations of explanatory factors (Black 1993). The results would be unreliable if the researcher chooses only to report the findings that are consistent with his (or her) prior belief, or selects only the model that seems to support his (or her) conclusion from many models tried. The reported results that seem significant could be just by chance with certain extent of data mining. Even if the researcher might only emphasis on the results that are consistent with his (or her) conclusion, data mining can be minimised by reporting all the results that he (or she) found from different models or time period (or both). Black (1993) mentioned that data mining worsens when many people are investigating similar problems. This is because only the most unusual or striking researches can be published and they might involve many false start and blind alley by nature, which means the statistical results that seem significant will be biased (ibid.
The ‘loyalists’ support Efficient Market Hypothesis (EMH) by pointing out the problem of data mining or market imperfections, the ‘revisionists’ support EMH with time-varying risk premiums, whereas the ‘heretics’ challenge EMH and believe that the market is irrational and the psychological factors could influence the pricing of securities, where the abnormal returns can be earned by using specific trading strategies. The rationalists and revisionists believe that the stock returns cannot be predicted and thus the superior performance of specific investment strategies, for example the momentum and contrarian trading, and value investing are not an anomaly. They suggest that the superior returns are a result by chance instead (Black 1993). Besides, they argue that the abnormal returns are the result of higher risk premium (Fama and French, 1996). For instance, the value stocks are risker than the growth stocks, and thus the higher risk from investing in value stocks has generated the superior profits. However, a significant abnormal return can still be found even after adjusting for risk (Chopra, Lakonishok and Ritter …show more content…
Data mining occurs when a researcher tries many approaches to do a particular study, such as different time periods, different models and different combinations of explanatory factors (Black 1993). The results would be unreliable if the researcher chooses only to report the findings that are consistent with his (or her) prior belief, or selects only the model that seems to support his (or her) conclusion from many models tried. The reported results that seem significant could be just by chance with certain extent of data mining. Even if the researcher might only emphasis on the results that are consistent with his (or her) conclusion, data mining can be minimised by reporting all the results that he (or she) found from different models or time period (or both). Black (1993) mentioned that data mining worsens when many people are investigating similar problems. This is because only the most unusual or striking researches can be published and they might involve many false start and blind alley by nature, which means the statistical results that seem significant will be biased (ibid.