|
PureBytes Links
Trading Reference Links
|
hi TJ,
I had been trying to do this ( previous post below)in a spreadsheet,
and have a few more ideas to add to it.
Having the option of only trading one e.g.$10,000 lot would not be
as great as having an option to trade many at the same time,
dependent on wheather or not more than 1 security has a buy on the
same day.
e.g., i have 1 sys test that gets an 80% win rate but has a very
low exposure.
the number of days in this test of the overall market in a 10 year
span would be similar to about 50 years if only 1 trade was going.
so the equivalent would be having 5 trades going in the 10 year
period to capture most trades.( of course this would not be true as
sometimes there may be over 10 buy signals or more on the same day).
so it may need to have a system that can vary dependent on the number
of trades at the time.
the buy and sells would be date dependent for a test like this,
so it would not be accurate to only choose 1 security if there may
be over 10 securities to choose from on that day.
there is a problem with this idea. it would need some type of order
to select which security will be bought when there are many on the
same day, so it would either need to be done by the, securities
alphabetical name or by the actual price .
an example would be if there are 10 trades going, call them,
a,b,c,d,e,f,g,h,i,j, and c,f,j, have a sell signal on the same
day,then there are 5 buy signals, the lowest price or the first in
alphabetical order would be placed in c, followed by f and j and 2
new groups k and L.
if all had sell sigs and there was only one buy then automatically
it would be placed in A.
after the test would be run, there would be the compounded total of
a series of trades for each group.(A to whatever).
another idea would be the option to put aside x amount of the profit
of each group.
if the profit that has been put aside equals $10,000 or x amount then
this could be used to start the trade of a new group.
if i have $50000 i can trade 5 groups simultaneously, if the unused
profit =$10000 i can now trade 6 groups. e.t.c.
so if my original system had a low drawdown and a high winning
percentage i would be able to trade more of the overlapping trades
that occur on the same day as time goes by as well as getting a very
realistic compounded total of each group.
cheers: john.
> ----- Original Message -----
> From: Steve Dugas
> To: amibroker@xxxx
> Sent: Saturday, July 13, 2002 7:34 AM
> Subject: [amibroker] TJ - Plug-in capabilities?
>
>
> Hi TJ,
>
> Hope you had a very nice vacation. Welcome back!
>
> My employer will pay for me to take classes, so I am thinking
about signing up for a class in object-oriented programming. I just
have a couple of questions regarding what types of testing it would
be possible to do in AB with scripts/plug-ins. What I would like to
do is this:
>
> On the 1st day of the backtest period, test all stocks in list
for my buy signals, evaluate the ones that pass and buy the one that
best meets my criteria. Later, when I get a sell signal, sell that
stock and buy a (probably) different one using the same test
procedure as before. So, for example, over a 2 year period, my test
would consist of buying and selling maybe 10 or 20 different stocks,
using the same system. My questions are:
>
> 1. Using a script or plug-in, would it be possible to do this now
in AB?
>
> 2. If not, any plans to add this capability in the future?
>
> 3. If not possible now, could I instead write a stand-alone .exe
which would access the AB database directly via Object Model or some
other method and do this? In this case, I guess I would not actually
be using AB itself, just its database. The idea would be to do this
type of multi-stock testing and get the results by creating my own
report. Thank you very much for any help or advice!
>
> Best Wishes,
>
> Steve
>
> Yahoo! Groups Sponsor
>
> Click here to find your contact lenses!
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of
Service.
|