One Sector at a Time
Thursday, January 15, 2009 at 11:12PM 
I used to test ranking systems against each stock sector using the ranking system performance page. That was before I came up with my new ranking system analyzer EXCEL spreadsheet. Now I can determine the SMS_Sharpe and SMS_Sortino ratios for each sector using the baseline universe as the Minimum Acceptable Risk (MAR). See last couple of posts for a more elaborate description.
Instead of testing 6 sets of stocks (5, 10, 20, 40, 80, 160 top-ranked stocks) I decided only to test 20 top-ranked stocks. This is because my data collection is not automated and would have involve 6 sets x 12 sectors = 72 backtests if I did all of the data collection. So I cheated a bit but I believe what I did is acceptable for the sector-based test. Here are the results:
|
Sector |
5 Year SMS_Sharpe |
5 Year SMS_Sortino |
|
BASICM |
1.76 |
2.87 |
|
CAPGDS |
2.09 |
3.62 |
|
CONGLO |
0.59 |
0.92 |
|
CYCLIC |
0.87 |
1.45 |
|
NONCYC |
1.85 |
3.13 |
|
ENERGY |
1.06 |
1.78 |
|
FINANC |
1.84 |
3.29 |
|
HEALTH |
1.94 |
3.40 |
|
SERVIC |
1.98 |
3.19 |
|
TECHNO |
1.76 |
2.95 |
|
TRANSP |
1.52 |
2.73 |
|
UTILIT |
1.89 |
3.32 |
The results are good except for conglomerates (CONGLO) and perhaps cyclic stocks. This is not unexpected. I generally have trouble with conglomerates and cyclical stocks are also hard on ranking systems.
After seeing these results I tried blocking
conglomerates and cyclic stocks from my trading system
simulation using the buy rule
(Sector != CONGLO) AND (Sector != CYCLIC)
The difference in performance for the modified system was insignificant so I will leave this sector block rule out of the system.
By the way, I restarted the portfolio on the weekend but it bought no stocks. Hopefully I have implemented the work-around correctly. If not then the port will NEVER buy any stocks. I guess I'll have to monitor the port regularly.
Steve
Reader Comments (2)
Nice blog. I am not too sure exactly how you come up with your analysis and quantitative studies but I am sure I will find out when looking deeper on your blog. Have you tried to run the same analysis on the japanese stock market. I would be really interested. I am actualyl following closely the japan market and I even have a blog on it http://thenikkei225investor.over-blog.com/
If you have a minute, I'd love to know what you think about it
Olivier
Hi Olivier - You have a really good blog! As for analysis, I use http://www.portfolio123.com for fundamentals on U.S. equities. Unfortunately I don't have any tools for analyzing Japanese stocks or way of trading them. I'm interested in learning more about the Japanese stock market though. Seems like they have already gone through the banking pains so they should be much better off than the rest of the world.
Steve