Research On Online Portfolio Strategies Based On Financial Anomalies

Since the efficient market theory put forward,it has always been the core of the theory of financial market and the cornerstone of modern finance theory.And many other finance theories,like the capital asset pricing model(CAPM),arbitrage pricing theory(APT)and the option pricing model(OPT),were established on the efficient market hypothesis.However,doubt and debate for the efficient market theory in theory and practice are uninterrupted,because there were a lot of empirical test on the market,and paradoxical phenomena of the efficient market theory were found.And they are known as "financial anomalies" which are different from efficient market theory.Appropriately take advantage of these anomalies can effectively improve the return of online portfolio strategies and promote the effectiveness of the market.In this paper,we put forward three online portfolio strategy models and algorithms by combining with the financial anomalies of mean reversion and momentum effects,machine learning and online learning technology,and took the empirical tests.The main work and innovation points of this paper are as follows.(1)We proposed an online portfolio strategy based on absolute price statistics.Borodin et al.put forward the heuristic algorithm of Anticor strategy based on the historical statistical relationship of relative prices between different stocks.As the history relative price just show the price change in amplitude during the periods,rather than the direct price change trend.So in this paper,we put forward an online portfolio based on the statistical relationship of absolute price of the stock,because the absolute price can show the direct price change trend as we can see.At last,we carried on the empirical test using data sets in the domestic market,the results show that the strategy that we put forward is better than the benchmark and the strategy based on the relative price,and has good robustness.(2)We proposed an online portfolio strategy of mean reversion based on the volume information weighted price.The reversal effect in financial anomalies is not only related to yield,but also related to the impact of trading volume.The reverse effect of securities usually appears after high volume trading.Based on this conclusion,this paper combined multi-phase volume and price information with online learning technology to design a new online portfolio strategy innovatively by establishing optimization model and algorithm.The result of the empirical research with domestic market data set shows that the volume information promote the profitability of strategy,and proves that the strategy we put forward is better than other strategies and has good robustness.(3)We proposed a momentum online portfolio strategy based on the disposition effect.The behavioral finance explanation of momentum effect indicates that the inference of prospect theory "disposition effect" leads to momentum effect.Based on that conclusion,this paper constructs an index of the current share price over the average price of the investors' position to design an innovative online portfolio strategy which is based on the rule that giving greater weight to the winner stock with respect to average price of the investors' position.The result of the empirical research with the data sets in the domestic market shows that the strategy is superior to other momentum strategy and benchmark performance,and has good robustness.