[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [amibroker] Re: Two Days after the Bullish Inflection



PureBytes Links

Trading Reference Links




<BLOCKQUOTE 
>
  
  <SPAN 
  >I have bitten my 
  tongue with the strategy talk that is posted where everyone seems to go ga-ga 
  over high percent returns over the last $B!D(B. er $B!D(B.6 months$B!D(Band some even test 
  back into 2002!!!!   No mention of drawdowns, little mention of 
  compound annual return, and much of it handled via very complex Excel 
  worksheets, rather than something like AB.   <SPAN 
  class=023373705-01102003><FONT face="Courier New" 
  color=#0000ff> <FONT face="Courier New" 
  color=#0000ff><SPAN 
  ><SPAN 
  class=023373705-01102003> 
<SPAN 
><SPAN 
class=023373705-01102003>agreed, TA in 
excel is no way to go. way too hard to build and test new ideas on a wide range 
of issues, plus can't imagine any of those big nasty sheets are 
error-free...
<SPAN 
><SPAN 
class=023373705-01102003><FONT face="Courier New" 
color=#0000ff> 
<BLOCKQUOTE 
>
  
  <SPAN 
  >As I am trying to 
  test some of these VV stock selections, I need to see how and if it is 
  possible to test back several years, at least to 1998.  I know I can buy 
  back data, but the backtesting system with VV (which is what you need to use 
  in order to test their parameters) is terrible and a backtest from 1998 would 
  surely put me to sleep.
  <SPAN 
  > 
  <SPAN 
  >Ken<SPAN 
  class=023373705-01102003><FONT face="Courier New" 
  color=#0000ff> 
<SPAN 
><SPAN 
class=023373705-01102003>well, it's true, the vv backtest facility is extremely 
un-automated, to put it kindly. what I do mostly is QuickTests. assemble a set 
of promising dates, and you can run a bunch of strategies you want to test 
against one date, then run them all against the next date, etc, and that goes 
pretty quickly. relatively quickly, I should say. it's actually most efficeint 
to work backwards from the most recent date, so vv doesn't keep resetting the 
end date to today, since it's now after the new start date. (not sure if that 
made sense written, but try it both directions, you'll see what I mean.) 

<SPAN 
><SPAN 
class=023373705-01102003> 
<SPAN 
><SPAN 
class=023373705-01102003>the thing QuickTests don't account for, or help you 
plan for, is handling a longer-term signal where some trades, both profitable 
and not, get stopped out. do you keep buying new stuff, right up to the opposing 
signal? leave those positions open? this is another aspect of timing and money 
management that needs to get sorted. FWIW, recent markets also make me question 
whether a timing plan that only knows about long and short can be right. surely 
there are times uncertain enough to make cash your best bet, but that's not 
typically part of the general VV mindset I think.
<SPAN 
><SPAN 
class=023373705-01102003> 
<SPAN 
><SPAN 
class=023373705-01102003>I also think that if the idea is to identify and trade 
in the direction of general market trends, VV is no way to work on that. reading 
dates off charts, eyballing tops and bottoms, just crazy. better to work in AB, 
and find methods that test well against a representative index, or watchlist 
average, etc, then go back to VV with those dates and test your strategies 
against them. 
<SPAN 
><SPAN 
class=023373705-01102003> 
<SPAN 
><SPAN 
class=023373705-01102003>that's been my main direction of inquiry for a while 
now, but I can't say I've found a holy grail of timing that feels like more than 
a curve fit. I've been wondering what'd happen if I just took some of the usual 
suspect indicators, on some usual suspect index, and did the QuickTest thing in 
VV with some known good strategies. the point isn't whether the timing does well 
buying indexes, it's how VV strategies do with it, and the best ones do 
seem pretty profitable. still hard to know you're not dealing with curve fits 
there too though...
<SPAN 
><SPAN 
class=023373705-01102003> 
<SPAN 
><SPAN 
class=023373705-01102003>do I maybe remember you from new england VV mtgs? if 
so, and you're coming thursday, I'll see you there.
<SPAN 
><SPAN 
class=023373705-01102003> 
<SPAN 
><SPAN 
class=023373705-01102003>I'm babbling on here, bed...
<SPAN 
><SPAN 
class=023373705-01102003> 
<SPAN 
><SPAN 
class=023373705-01102003>dave






Yahoo! Groups Sponsor


  ADVERTISEMENT<img
src="" width=300 height=250 border=0>  









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 the Yahoo! Terms of Service.