|
PureBytes Links
Trading Reference Links
|
Hi b519b,
A. I have downloaded from
http://quotes.nasdaq.com/quote.dll?page=nasdaq100
the 101 stocks of ^NDX(100+^NDX) from Jan 2000.
B. Only PALM entered at 2 Mar 2000, all the others cover
that period.
C. I have constructed, based on the official N100 above
catalog, the N100.tls file, to download data directly with
Amiquote.
D. All composite tickers referred in AFL Library are
caculated for ^NDX for the period Jan 2000-July 2001
E. In Yahoo download there exist some holes and some
splits are missing
F. What is the quantity you calculate in spreadsheet ?
If any of above may help you, let me know.
Dimitris Tsokakis
--- In amibroker@xxxx, b519b@xxxx wrote:
> I should have been a bit clearer in my first post. Thanks to
Peter's
> comments, I can make a couple clarifications:
>
> > The first problem I see is that QQQ didn't start trading until
> early 1999 so
> > your study will not work with QQQ if you go back to 1985.
>
> I've already done the back testing of the QQQ from its start in
March
> 1999 using an indicator batch of about 50 stocks. That was done
with
> a spreadsheet. The results are encouraging but there was one
drawdown
> of over 30% in the two and half years of QQQ history. If such a
> drawdown only occurs every couple of years, the method may be
> workable, but not if it occurs every year. So I want to use the
> Nasdaq 100 index as a proxy for the QQQ so to get 10 to 15 year
test
> period to see how often the indicator might fail miserably.
> Comparison of the proxy results and those for the QQQ results for
the
> overlapping period (March 1999 to the present) should confirm that
> the Nasdaq 100 is a reasonable proxy for the longer test.
>
> > I think what you would want to do is make a new composite
> > ticker with a list of stocks. There is a newsletter that
explains
> how to do
> > this.
>
> No, I really don't want a composite ticker of a new list of stocks.
> That would only measure the variations of the list as a group,
> but I want to measure the variation of each stock and then total up
> the individual variations (+1 if it is positive no matter how
large,
> or -1 if negative). Since the indicator is based on the direction
of
> the moves not their magnitude, a composite of the stocks would not
> work.
>
> What I am trying to get is an overbought/oversold indicator based
on
> the movement of individual stocks. My reference to a ticker in the
> pseudo code probably confused matters on this. Given my limited
> undestanding of AFL, the only way I can see to store the results of
> indicator calculations would be to put them into a stock like data
> array. Amibroker would think it was a stock, but I'd know that that
> the data in the Open, High, Low, Close, etc were aspects of the
> indicator and program the buy/sell conditions according.
>
> Why am I using a work around of storing indicator results in a
stock
> like data list? Good question. I'm not sure if I need to, but
> it would save Amibroker from having to recalculate a rather long
> and set of calculations for multiple back tests. Perhaps it would
> best to just let Amibroker recalculate the indicator fresh each
time.
> Or maybe, I should just go back to an Excel spreadsheet to create
the
> indicator. It would then be a simple task to have the spreadsheet
> results saved as in a stock like format and imported when needed
into
> Amibroker using the foreign call feature.
>
> Since it took a couple days to do a 50 stock indicator for two and
a
> half years in the spreadsheet, 10 years plus might take a week or
> two. I was hoping that with the right coding Amibroker might do
this
> in less than a day. Any comments?
|