|
PureBytes Links
Trading Reference Links
|
hmmm. this was an assumption on my part, one that seemed logical, and still
does, to me. but your questioning of it, from your vastly more experienced
viewpoint, made me do some actual checking, in progress as I write. you do
mention it being a bigger issue before 3/00, and usually the first step of
my basic testing runs 3/1/98 to 3/31/03, so maybe we're not saying anything
contradictory anyway.
logically, membership in the NASDAQ 100 or S&P 100 5 years later seems like
it must be a very strong future-oriented filter of some kind. maybe not to
the long side per se, but that's a pretty specifically selected group of
issues, using a criteria based 5 years in the future. even with a
tradeableness filter applied to both data sets, logically that fails the
sniff test for a neutral sample.
mark, can you explain to me why this wouldn't be so? aside from any actual
observed results?
at the very least, if we trade strategies on this universe in real time,
we're not executing what we tested, since we have only the current
memberhips available to trade, not the ones from 2007. if the
future-index-members-filter results are in fact comparable to real time
results on all stocks, for many kinds of strategies, that would strike me as
somewhat accidental.
but that's theory, and we know where that gets us... some actual numbers are
below.
----------------------------
DEFINITIONS
----------------------------
"N+S" is NASDAQ 100 plus S&P 100. I don't completely trust that my current
watchlists for these are exactly accurate, given recent changes.
"liquid" means LLV(v * c, 50) > 1000000. only stocks with at least one
liquid bar in the test range are included in any of these tests.
"% liquid of bars in test" is the percentage of all populated bars that were
liquid by that definition, averaged over all included stocks.
"% liquid and up of bars in test" is the percentage of all populated bars in
range that were liquid and closed higher than the previous day, averaged
over all included stocks.
"% up of liquid bars" is the percentage of liquid bars in range that closed
higher than the previous day, averaged over all included stocks.
"all stocks" is QuotesPlus current data.
all figures rounded to 2 decimal places.
=======================================
% liquid of bars in test
----------------------------
N+S all stocks
----------------------------
1998 90.96% 60.98%
2002 100.00% 70.94%
----------------------------
% up of liquid bars
----------------------------
N+S all stocks
----------------------------
1998 49.21% 42.49%
2002 45.24% 45.77%
----------------------------
% liquid and up of bars in test
----------------------------
N+S all stocks
----------------------------
1998 44.85% 28.40%
2002 45.24% 33.44%
----------------------------
number of qualifying stocks
----------------------------
N+S all stocks
----------------------------
1998 181 746
2002 190 1430
----------------------------
=======================================
not sure what of this is statistically significant, but...
- not too surprisingly, fewer bars were liquid in 98 than in 02. the effect
is somewhat stronger in all stocks than N+S, but not radically so.
- in 02, an almost identical fraction of liquid bars were up for N+S and all
stocks. in 98, somewhat MORE than that were up for N+S, and FEWER were up
for all stocks. long future bias and volume inflation respectively, maybe?
- the similarity between '98 and '02 in % up of liquid bars is surprising to
me, since '98 "looks" up and '02 "looks" down on most indexes.
dave
> Well, IMO, it is absolutely not necessary to test against "all
> stocks." Only a representative sample. And oh, BTW, the stocks of
> some indices are more representative than random samples, not to
> mention more tradeable. I hear you on the looking into the future
> argument but that was more applicable before Mar 00. I don't see a 5
> year "major long bias" when I scroll through the ND100 stocks. But I
> do see a lot of bear.
>
> --- In amibroker@xxxxxxxxxxxxxxx, "Dave Merrill" <dmerrill@xxxx>
> wrote:
> > not chande, don't play him on tv, but here's my take fwiw. IMO, the
> > selection of the initial universe to test is the result of another
> set of
> > filters with another set of parameters that aren't being talked
> about at the
> > moment, and are just as subject to (over)optimization as anything
> else.
> >
> > most fixed universes smaller than All Stocks are really hard to
> backtest
> > reliably without bias, so I don't when I'm testing for real. for
> example,
> > testing in '98 using the '03 N100 is looking into the future and
> selecting
> > only stocks that ended up big 5 years later -- a major long bias.
> even if
> > you have historical N100 membership data, for example, it's hard to
> apply it
> > on a continuous basis over the life of a test.
> >
> > so, I often test ideas out on small subsets for speed, but nearly
> always
> > plan on running against all stocks eventually, and build in the
> criteria
> > needed to narrow the stocks considered as required. if you do
> anything else,
> > I'd suggest it needs some careful thought.
> >
> > dave
------------------------ Yahoo! Groups Sponsor ---------------------~-->
Buy Ink Cartridges or Refill Kits for your HP, Epson, Canon or Lexmark
Printer at MyInks.com. Free s/h on orders $50 or more to the US & Canada.
http://www.c1tracking.com/l.asp?cid=5511
http://us.click.yahoo.com/mOAaAA/3exGAA/qnsNAA/GHeqlB/TM
---------------------------------------------------------------------~->
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
|