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

RE: Objective functions (was RE: [amibroker] Re: Optimization -- again) - to Fred



PureBytes Links

Trading Reference Links




<SPAN 
class=016265511-20102003>come on fred, I'm serious, you know what I'm 
asking.
<SPAN 
class=016265511-20102003> 
<SPAN 
class=016265511-20102003>we don't need to know specifically how far up something 
will go, that's just one way of knowing something about when it will go back 
down.
<SPAN 
class=016265511-20102003> 
<SPAN 
class=016265511-20102003>the point is to profit for our trades. that requires 
buying lower than we sell. what aspects of market behavior have been constant 
over time that enable us to do that?
<SPAN 
class=016265511-20102003> 
<SPAN 
class=016265511-20102003>dave
<BLOCKQUOTE 
>In 
  order to make money, which is what I thought the goal was, do we really 
  need to know how high is up etc. ?  --- In 
  amibroker@xxxxxxxxxxxxxxx, "Dave Merrill" <dmerrill@xxxx> 
  wrote:> knowing that the markets go up and down isn't tradable 
  knowledge. you have> to know something about *which* of those two 
  things will happen, to which> stocks, when, and/or how far. as far 
  as I know, patterns on those levels do> change over time, or at 
  least the lengths over which they cycle change.> > what 
  *tradable* market behaviors are there that are constant over time?> 
  > dave>   From: Fred [mailto:fctonetti@xxxx]> 
  >   Uhhh ... the ups and the downs ... as far as I can tell 
  marlets have>   pretty much done that since the beginning 
  of time.  Nothing much>   different about it in my view 
  today the it was in any other time>   frame.> 
  >   --- In amibroker@xxxxxxxxxxxxxxx, "Dave Merrill" 
  <dmerrill@xxxx>>   wrote:>   > I'm 
  serious fred. what kinds of tradable market behaviors are 
  you>   talking>   > about that aren't 
  related to things that change over time?>   
  >>   > basic example: virtually every description of 
  market behavior I'm>   aware of>   > 
  has time constants, trigger levels, and other "static" 
  features>   whose best>   > performing 
  values migrate or cycle over time. it seems unlikely 
  on>   the face>   > of it that the 
  point where some specific MA crosses another>   specific MA 
  is a>   > quasi-permanently useful switch point, for 
  instance. what inherent>   mechanism>   
  > of market behavior that makes this optimum, as opposed to 
  some>   other pair of>   > MAs? is it 
  really possible that these specific parameter values 
  are>   > constant, given all the changes in the 
  economy, the trading>   population,>   > 
  analysis technology, etc?>   >>   > 
  you must be talking about some other level of behavior 
  that's>   constant in>   > some 
  pan-historical sense, but I'm lost without an example of a>   
  tradable>   > feature like this.>   
  >>   > (it's interesting to me that auto-optimizing 
  system don't have>   those kinds>   > of 
  static parameters in the same sense. yes, they have specifics 
  of>   course,>   > like constraints on 
  the range of each parameter, time constants on>   
  their>   > learning behaviors, and a definition of an 
  equity metric. but they>   make no>   
  > assumptions about what time constants or crossover levels 
  work>   well, they>   > just try 'em and 
  see.)>   >>   > 
  dave>   >>   >   forest 
  (:-)>   >>   >   what kinds 
  of tradable market behavior should we be looking>   at/for 
  that>   > transcend the "short-sighted view of history" we 
  *shouldn't* be>   looking for?>   
  >>   >   dave>   
  >>   >     This makes me want to 
  ask what your longest possible time frame>   is 
  ?>   >>   >     
  --- In amibroker@xxxxxxxxxxxxxxx, "Dave Merrill" 
  <dmerrill@xxxx>>   >     
  wrote:>   >     > well yes, you're 
  right, the same stuff is always happening.>   
  prices>   >     go 
  up,>   >     > prices go down, and 
  they always have.>   >     
  >>   >     > but that's not 
  useful info to trade on. what we care about is>   
  >     trends of some>   
  >     > kind that can be predicted/hoped to continue 
  or reverse in>   some>   
  >     particular>   
  >     > time frame. that's knowledge we can profit 
  from. and those>   trends>   
  >     come and>   
  >     > go constantly, on every time scale. these 
  shorter-term moves>   are>   
  >     what we>   
  >     > trade.>   
  >     >>   
  >     > here's my question I guess: if I only see 
  behavior that never>   >     
  changes over>   >     > the longest 
  possible time frame, what do I see that I can use?>   
  >     >>   
  >     > dave>   
  >     >   There are a lot of questions and 
  provacative statements in>   your>   
  >     post,>   
  >     >   only one of which from my 
  perspective needs an>   answer/response.>   
  >     >>   
  >     >   Market behavior will continually 
  change after that ...>   >     
  >>   >     >   Change 
  ? from what ? into what ? I guess this is the part I>   
  don't>   >     >   
  follow.  To me there is nothing new in market behavior 
  now>   that>   
  >     >   didn't exist last month, last 
  year, last decade, last>   century, but>   
  >     >   clearly those that take a short 
  sighted view of history and>   the>   
  >     >   market action that made up that 
  history will clearly never>   see it.>   
  >     >   It's a forest and trees thing 
  ...






Yahoo! Groups Sponsor


  ADVERTISEMENT 









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.