<<I always beat the S&P or I wouldn't do it for a living. My money never stays sterile.>>
Then I have to admire you investment skills. I know I have
been unable to do this. That is why I seek a mechanical basis.
<<We are in the biggest bull of all time.>>
I want to be able to expect to do well (better than the SPX)
whether the market continues strong or otherwise. That suggests
a partly hedged position to me.
<<Testing should include as many years as you can
muster not just a year or two.>>
I totally agree. Unfortunately I see no feasible way to get the
data (price, volume and funny) in the 'as it was when traded' form.
So, I will slowly augment my history over time.
<<As I said, I can not make sense of your tables or varables,
I still waiting for bottom line. Do you think that if you buy the long stocks, you should
expect a 65% return?>>
I am surely not saying I expect a 65% annual return for the
highest long stocks selected. All I have is 13 months experience,
in a time frame where the SPX did somewhere in the low twenties.
All I am saying is the variables I am using long far outperformed the
SPX for virtually all time segments of 3 months or more during this
time frame. And being short with the 'worst' stocks did pretty
good as hedges. What more could you say if work you were doing
produced equivalent results?
<<Are you saying just buy any thing on the long list and hold it for 6
months before selling? Is that your entry/exit plan? then where does money go? into
your next list.>>
I am thinking of being long the best 4-6 stocks and holding
for say 6 months, and then selling and repeating the process
unless I find reason to do otherwise. Incidentally, during the
past year, some stocks like DELL MSFT AOL as well as some nontech
stocks remained within the top 10 for many of the 13 months,
so 100% turnover did not have to occur.
I'm not sure yet about the short side.
<<The past is important to us. In the 56 variables there might be one or two that count for
90% of the correlation to profit, most would have little relationship.>>
This is not the case, at least within the 13 months experience.
After starting with a core group, variables were not added unless
their impact on return with and without that variable appeared
material. I have run many, many 4-12 variable filters trying to get equivalent returns to that for the full set of 50+. I couldn't.
Some might perform well for parts of the time frame, but no small
variable set was equal to the full set over the full time frame.
<<Everything I say is my opinion as it should be>>
Same here. |
