Research On Barra Factor Stock Selection Model Based On Quantile Regression

Now quantitative investment has become one of the most important investrmen-tmethods in Western financial markets.In the fut.ure,quantitative investment will also occupy an important position in China's financial market.With the ad-vancement of China's capital market reform,the strengthening of the liquidity of financial products and the continuous enrichment of financial derivatives,quan-titative investment will play an increasingly important role in China's financial sector.Stocks are one of t.he most important invest.ment t,ypes in the financial market.Therefore,multi-factor,stock select.ion is a quantitative strategy that is of primary concern to doImestic ancd foreign investors.The three-factor model and the five-factor model of Fama and French arc the milest.ones in the multi-factor stock selection model.Since then,a large number of scholars ancd investors have conducted in-depth research on multi-factor selection and model construction.In the field of quantitative investment,the selection of effective factors for mult.i-factor models mainly uses scoring and regression.This paper uses the re-gression method and builds a port.folio based on it.This pa.per selects the Barra style factor based on the Barra factor model developed by China's stock market.Since the factor value distribution of each stock does not obey the normal distri-bution,most of them have the problem of peak and thick tail and extreme value,so we choose the quantile regression.More suitable than linear re.gression.First,using the unary quantile regression to verify t.he validity of the factor,then use the correlation test to eliminate the redundant,factor and select five factors;final-ly,based on the factor to the tail of the high-yield stock and the low-yield stock Influence,using the multivariate quantile regression model to construct a factor selection strategy,using the current,factor value and t.he regression coefficient to calculate the expected rate of return,the stocks are grouped and backtested.The results show that from the perspective of the absolute value of the re.gres-sion coefficient of the stock factor,the nonlinear market,capitalization fact.or and growth fa.ctor will have a relatively significant impact on high-yield stocks;while the scale,momentum and volatility factors affect the low-yield stocks.It,will be noticea.ble.For the quantile regression multi-factor strategy constructed by the growth factor and the nonlinear market capitalization factor,the annualize.d rate of return during the sample period of the 90%quantile is 29.55%,which is higher than the year of the QR(0.1)strategy.Rate of return;for multi-factor strategies constructed by size,reversal,and volatility factors,the annualized rate of return during the sample period at the 10%quantile is 39.26%,which is higher than the QR(0.9)strategy.The annualized rate of return,in addition,the annua.lized rate of return of the two models is higher than the annualized rate of return of the benchmark model strategy;at the same time,it also demonstrates that in the Chinese stock market,the scale factor,the reversal factor and the volatility factor are more Strong applicability.