|
PureBytes Links
Trading Reference Links
|
This technique is not only applicable to evaluating robustness, but
also to issue selection and is simple but I believe very powerful. It
addresses stationarity. *Very roughly*, a time series is stationary if
the rules that generate it don't change over time and thus the
statistical parameters of the process don't change with time. There
are actually degrees of stationarity and statistical tests for it
ranging from simple to arcane, but as I said in the first post we're
trying to cut through the crap. So instead of statistically analyzing
the *data* to see if they're stationary and then wringing our hands
*when they're not*, we're going to look at stationarity with a twist:
the stationarity (consistency) of *how our system trades the data* and
we're going to do it with pictures!
To select which stocks to look at, go back into AB and perform the
same test you did for criterion #2 because we want to look at the most
actively traded issues again. You'll be doing this 3 times, once for
each category of stock. Set AA range setting on a range that
contains a bull, bear and sideways period (Mar 98- Mar03 is OK),
commission setting at least .5% and no trades in the reports tab,
backtest system on all stocks in the group. Now sort by # trades
(with most on top). Look at the PFs and write down the names of any 5
stocks (you should actually do as many as you can stand) with decent
PFs (depends on the system and avoid the best performing for now).
Reset AA (apply to current stock, range all quotations, and settings
*with* trade list). Now backtest the first stock on the list. Copy
and paste the results into a spreadsheet. Sort by column b(trades)
and *delete all "out" rows*. Re-sort (if nessary) by date so trades
are listed from earliest to latest. Now highlight only the headers
and cells with data in them (you'll get better graphs than if you
highlight the whole column) in the %profit and profit/bar colums and
graph them separately as column charts. Looking for a fair degree of
consistency in *both* % profit per trade and % profit per bar on *all
15 stocks*. Now what precisely that means could fill a library but
experience is the best teacher so try it out on as many systems and
stocks as you can stand and you'll start to get a feel for it. Two
important questions to ask yourself while looking at these are, could
I handle being on the receiving end of any part of this sequence, and
does this look so consistent that we're likely to see more of the
same. Sorry for the nebulosity but if you graph enough of these, you
*will* get a feel for it. Also, we can apply some more advanced
techniques (such as control charts) later if you like but that
requires special software and I think eyeballing is more than
sufficient, especially for now.
Worst case is this will give you an interesting view of how your
systems trade. If any of you don't like it, I'll cheerfully refund
your money, to include postage! Will post some examples as soon as I
can.
------------------------ 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/
|