1. Summary
In the previous Barra model series articles, we constructed the Size factor, Beta factor, Momentum factor, Residual Volatility factor, NonLinear Size factor, Book-to-Price factor and Liquidity factor, and created corresponding single-factor strategies respectively, among which The Size factor and the NonLinear Siz factor have strong earning power. This section of the article is the eighth in the series, and the Earning_Yeild factor will be further constructed under this series.
2. Model theory
The Earning_Yeild factor of the Barra model is calculated as follows:
The Earning_Yeild factor is a composite factor consisting of three sub-factors, namely: profit forecast, cash flow ratio and historical profit-to-market ratio, with weights of 0.68, 0.21, and 0.1. Earnings forecast data is the data expected by analysts. It has not been provided yet, and it will be added in the future when it is provided. However, the proportion of earnings forecast is as high as 60%, and the entire Earning_Yeild factor will be quite different from expectations.
3. Factor Analysis
Use alphalens to analyze the Earning_Yeild factor (2022-March 30, 2023).
Judging from the above income analysis, the alpha income under each rebalancing cycle is almost 0; both the largest group and the smallest group of Liquidity factor values contribute positive returns, and the positive income of the smallest group is much greater than that of the largest group.
Further, from the perspective of information coefficient, the average IC and standard deviation of IC are not much different under each rebalancing frequency, and the average IC is as low as 0.02, which is worse than the Liquidity factor in the previous period, and the stock selection ability should not exist.
From the group income graph, the Earning_Yeild factor presents positive returns at both ends and negative returns in the middle. But from the perspective of factor logic, the Earning_Yeild factor represents profitability. The higher the profitability, the higher the return of the stock should be, which is contrary to the law shown in the above-mentioned factor return distribution diagram.
The main reason may still be the lack of sub-factors (earnings forecasts), but due to lack of data, the construction of the Earning_Yeild factor is not complete. However, in order to further comprehensively analyze the 10 factors of the Barra model in the later stage, we will further write the code here for later use.
4. Backtest analysis
Backtesting time: 2022-01-01 to 2023-03-30 (share exchange on the last trading day of each week)
Backtest species: All A shares (excluding ST shares, suspended stocks and sub-new shares within one year)
Initial capital: 1 million
Handling fee: 0.0007 (bilateral 10,000 commission + unilateral 1,000 stamp duty, a total of 1.4 thousand, that is, bilateral 10,000 7)
Slippage: 0.00123 (two sides thousand 1.23)
Maximum number of positions: 30
During the backtest period, the strategy achieved an annualized rate of return of -3.95%, a maximum drawdown of 20.33%, and a Sharpe ratio of -0.19. The performance of the overall strategy is poor, which is in line with the above-mentioned factor analysis results; the net value of the strategy is highly consistent with the market trend, but the fluctuation is relatively small.
Above, the strategy source code of this issue has been shared with the Nuggets Quantitative Community, and you can obtain it yourself through the link below.
Portal: https://bbs.myquant.cn/thread/3416
References for this issue:
1.The Barra China Equity Model (CNE5) - MSCI - MSCI
Link: https://www.docin.com/p-1377763566.html
2. Ishikawa correctly understands Barra's pure factor model
Link: https://zhuanlan.zhihu.com/p/38280638
3. Preliminary research on the Barra model of Founder Securities, analysis of A-share market style
4. BOC Securities' thoughts on Barra's China equity CNE5 model
5. Quantitative Investment Notes Barra Series
Link: https://zhuanlan.zhihu.com/p/68110181
6. Understand the nonlinear market value factor NLSIZE/MIDCAP
Link: https://zhuanlan.zhihu.com/p/150310851?from_voters_page=true