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

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



PureBytes Links

Trading Reference Links




<SPAN 
class=311242611-20102003>Pal, couple questions/comments.
<SPAN 
class=311242611-20102003> 
<SPAN 
class=311242611-20102003>- are you saying that 30 "occurrences" in any system 
produces 95% accuracy? 30 trades? regardless of the market or trading system 
rules or time frame? what's the basis for saying this?
<SPAN 
class=311242611-20102003> 
<SPAN 
class=311242611-20102003>- could you explain "select stable parameters with an 
equity shift less than the parameter shift after equity spikes have been 
eliminated"? I don't understand what you mean.
<SPAN 
class=311242611-20102003> 
<SPAN 
class=311242611-20102003>- just fyi, your last paragraph seems to be trying to 
convince me that optimizing is good, probably in response to my asking "if you 
prefer not to optimize parameters, how do you set them?". I asked that only 
because you said, "I prefer a system to work without optimization", which I 
thought was a nice goal, but one I don't understand how to achieve. seems that 
you don't actually intend to avoid optimization either, since you then discuss 
how you do it.
<SPAN 
class=311242611-20102003>dave
<BLOCKQUOTE 
>There 
  is a correct method to optimize any system that is statistically valid, 30 
  occurrences with 95% accuracy.The key to optimization is to select 
  stable parameters with an equity shift less than the parameter shift after 
  equity spikes have been eliminated. This process creates stability for 
  optimal parameters shifts within the four technical market phases. 
  Parameter shift is always geometric, but equity shift decline relative to 
  unstable parameter selection is usually exponential.All systems 
  are optimized to some degree. As soon as a trader chooses to enter a trade 
  on the open as opposed to the high/low/close of day, he has made a 
  decision as to how a system should be traded. Does he know the 
  high/low/close of day entry is better than the next opening for an entry? 
  If not why not? A potential 28% difference in profitability exists for 
  channel system entries between opens and closes.The purpose of 
  trading is to consistently make money. This is done by having the best 
  information available. If a trader does not know the best entry for his 
  system, what is he trying to prove? That the system isn't optimized? To 
  lose money because a trader is ignorant of his systen's best parameters is 
  foolish.Regards,Pal--- In amibroker@xxxxxxxxxxxxxxx, "Dave 
  Merrill" <dmerrill@xxxx> wrote:> one question pal: if you 
  prefer not to optimize parameters, how do you set> them? or do you 
  have some kind of trading rules that don't have time> constants, 
  trigger levels, etc, that need to be set?> > dave> 
  >   I thought I might throw in my 2 cents.> 
  >   Vendors love optimization, because it can generate eye 
  popping>   hypothetical profits which has no connection to 
  real-time trading.> >   I prefer a system to work 
  without optimization. But if I have to do>   it, I would 
  make sure that the optimization is robust in the>   following 
  manner:> >   1. The sample size of data should be 
  large enough to represent real->   time market conditions 
  - bull, bear and sideways markets.> >   2. The 
  look-back period should be as large as possible for the 
  same>   reasons.> >   3. The 
  testing of optimizable parameters should be on out of 
  sample>   data using walk-forward analysis.> 
  >   4. The Central Limit Theorem says that for a sample to 
  assume the>   characteristics of the population, the size of 
  sample should be>   large. The minimum sample size should be 
  around 30. But since an>   uptrend or downtrend can last for 
  say 50 periods, I would have a>   minimum sample size of 100 
  periods making sure that the full market>   cycle is 
  there (uptrend, downtrend and congestion).> >   5. The 
  optimizable parameters should be as few as possible and 
  tested>   in a wide variety of markets.> 
  >   Curve-fitting is like rolling a fair dice with 1/6 
  probability of>   getting any number from 1 to 6, rolling it 
  5 times, getting #6, 4 out>   of 5 times (80%) of 
  time.> >   A lot of traders fall in the trap of 
  curve-fitting without being>   aware of it. So when designing 
  a system, it is important to keep your>   guard up as far 
  as curve-fitting is concerned.> >   Regards,> 
  >   Pal>   --- In amibroker@xxxxxxxxxxxxxxx, 
  "Gary A. Serkhoshian">   <serkhoshian777@xxxx> 
  wrote:>   > Fred,>   
  >>   > Could you narrow-down your idea of a reasonable 
  sample size for>   backtests.  You've been hinting at 
  rather sizeable backtesting>   periods, but would like to put 
  some numbers to it.  Also wonder if>   you use # of 
  trades as a guide versus period of time for 
  backtesting>   period.>   
  >>   > Thanks,>   > 
  Gary>   >>   > Fred 
  <fctonetti@xxxx> wrote:>   > 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 ...>   >>   > --- In 
  amibroker@xxxxxxxxxxxxxxx, "Dave Merrill" 
  <dmerrill@xxxx>>   > wrote:>   > 
  > I'm not trying to be argumentative, honest (:-)... I'm more 
  than>   a>   > 
  little>   > > sick of saying the same thing over and 
  over, but I  j u s t   d o>   > n ' 
  t   g>   > > e t   i t 
  .>   > >>   > > 
  ------------------------------>   > 
  >>   > > I fail to see the huge difference in 
  principle between equity>   > feedback 
  and>   > > backtesting.>   > 
  >>   > > let's start by assuming that backtesting 
  performance of a system>   > and 
  its>   > > parameters over some period of past data 
  tells you something>   about>   > 
  its>   > > future performance. it's not a perfect 
  predictor, but it's the>   best>   > 
  evidence>   > > we have. does this seem like a 
  reasonable starting point? what>   > 
  alternative>   > > is there?>   > 
  >>   > > if that's true, why is it better to do it 
  only once? what>   > justification is>   
  > > there for picking one examination period over another? 
  clearly>   > market>   > > behavior 
  will change continually after that. don't we need a 
  way>   of>   > 
  working>   > > that looks at what's been happening and 
  evolves our response?>   > >>   > 
  > sounds like we examine performance up to some point and 
  adjust,>   > trade with>   > > 
  the best-choice system and parameters for a while, then 
  examine>   and>   > 
  adjust>   > > again later. make sense? what alternative 
  is there?>   > >>   > > so then, 
  how often do we re-examine performance history? to put 
  it>   > > differently, how long do we ignore any 
  changes in market dynamics>   > that 
  may>   > > or may not have occurred? why would 
  intermittently refusing to>   look>   > 
  and>   > > respond improve system performance or 
  reliability?>   > >>   > > if 
  that needs to be done, why not have the system itself do 
  it,>   as>   > part 
  of>   > > its inherent operation? why is it better for 
  us as an outside>   agent>   > 
  to>   > > periodically run some separate tests, reach 
  into the internals of>   > the>   
  > > system, and change stuff?>   > 
  >>   > > or should we just continue with the system 
  and parameters we>   choose>   > at 
  the>   > > beginning? are they somehow more valid than 
  what we'd choose>   later,>   > 
  using>   > > the same backtesting methods, but on a 
  different date range of>   data?>   > 
  >>   > > 
  ------------------------------>   > 
  >>   > > I realize that even if it seems to make 
  sense logically, this all>   a>   > 
  complete>   > > crock if no systems put together like 
  this even backtest well,>   > never 
  mind>   > > forward testing.>   > 
  >>   > > but every time I think about abandoning 
  this line of research, it>   > seems 
  like>   > > the first thing I'd want to do with a new 
  system would be (let me>   > 
  guess),>   > > test and possibly adjust it using data 
  up to some date, then run>   > with it 
  for>   > > a while after that and see if equity growth 
  is good. if it is,>   I'd>   > want 
  to>   > > lather, rinse and repeat with other in and 
  out of sample data, to>   > make 
  sure>   > > that wasn't 
  coincidence.>   > >>   > > 
  sounds way too familiar to be a completely different 
  animal.>   > >>   > > 
  dave>   > >   From: Fred 
  [mailto:fctonetti@xxxx]>   > >>   > 
  >   That IS what I was trying to say.  I suspect because 
  equity>   feed>   > 
  back>   > >   is like looking in a rear view 
  mirror, great for letting us know>   > 
  >   where we were and how we could have adjusted the past to 
  make it>   > >   better, but that's about 
  it.






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.