The in-the-sample period starts from 2009-6-1 to 2009-12-31. Basing on the data of 2009, three methods are used for simulating the portfolio. Table 8 lists the statistics summary comparison between the in-the-sample period and out-of-sample period (2010.1.1 to 2010.8.13). The left part of table is the statistics summary at simulations point while the right part depicts the result after 8 months. From the table it is found that the simulations of these three methods are very successful and there are only small changes of all these indicators for these three methods after 8 months. Among these three methods, correlation method performs the best and after 8 months, the correlation coefficient could still remains 0.9946 and track error is only 0.13%.
Summary of Portfolio at Simulation Point |
Summary of Portfolio after 8 months | ||
Tracking Error Method |
Tracking Error Method | ||
Tracking Error |
0.09% |
Tracking Error |
0.13% |
Correlation Coefficient |
99.86% |
Correlation Coefficient |
99.50% |
Beta |
1.00 |
Beta |
0.98 |
Correlation Method |
Correlation Method | ||
Tracking Error |
0.10% |
Tracking Error |
0.13% |
Correlation Coefficient |
99.86% |
Correlation Coefficient |
99.46% |
Beta |
1.02 |
Beta |
0.99 |
Beta Method |
Beta Method | ||
Tracking Error |
0.16% |
Tracking Error |
0.13% |
Correlation Coefficient |
99.57% |
Correlation Coefficient |
99.51% |
Beta |
1.00 |
Beta |
0.98 |
Table 8
It is also found that the similar conclusion on the basis of performance comparison between Simulated Portfolio and Benchmark. Table 9 provides the performance of CSI 300, the portfolio return for these three methods as well as 8 mainstream index funds in Chinese stock market. From the table it is found that the simulation method of correlations performs the best among three simulations methods and it gets the closest return comparing with the CSI 300 in the out-of-sample period. The beta method gets the highest absolute return but from the aspect of index fund construction it performs the worst because it gets the biggest deviation from the benchmark. Finally if comparing the performance of Correlation method with that of the mainstream index funds, it also gets the best performance with the smallest deviation with the CSI 300(benchmark) comparing with other ETF funds.
Summary of performance between Simulated Portfolio and Benchmark
Name |
2010-1-1 to 2010-8-13 |
CSI 300 Return |
-20.14% |
Portfolio Return for Tracking Error Method |
-19.62% |
Portfolio Return for Correlation Method |
-20.10% |
Portfolio Return for Beta Method |
-18.64% |
100ETF |
-16.94% |
Huaan 180ETF |
-20.78% |
Jiaoying 180ETF |
-22.30% |
Dividend ETF |
-23.00% |
Gongying 50ETF |
-25.12% |
Huaxia 50ETF |
-23.37% |
Nanfang ETF |
-7.98% |
Boshi ETF |
-20.33% |
Table 9
Finally provided is the cumulative return series comparison between these three portfolios and CSI 300(market portfolio) in Diagram 16. It is shown that as for these three simulation methods, they are all very effective for simulating the CSI 300 index and all of them could maintain a very small deviation from the index even after 8 months since the simulation date.
Although Chinese Stock market is not as efficient as that in developed countries and information disclosure is also not so fair and open, the passive quantitative portfolio construction is quite effective in Chinese A-share stock market. So far the empirical result of passive quantitative investment is more encouraging than that of the active quantitative investment; The above phenomenon is partly due to the backward in IT application in Finance and most of index fund quantitative analysts in China could not write the computer codes and either rely largely on some so-called “blackbox” software or are still exploring the effective simulation method according to the recent interview with 8 ETF fund companies in China.
评论