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

Re: [amibroker] Re: solving the prediction problem



PureBytes Links

Trading Reference Links




Pal, you said earlier "...Statistics is meaningless in the context that you 
can come out with statistics to disprove anything." So, I misspoke. You were 
talking about disproof, whereas I was speaking of proof (mathematically, one is 
1 minus the other). But no matter. Let me go one step further. Suppose the 
null hypothesis is A is not equal to B. Put in terms of trading parlance, the 
null hypothesis might be when the short MA of C crosses above the long MA of C, 
prices will not rise. In order to disprove that hypothesis, I have to try to 
prove that A = B (the alternate hypothesis); in other words, I have to prove 
that when the short MA of C crosses above the long MA of C, the price rises. So, 
I design an experiment (a backtest) to get as much information to prove that A = 
B. If I obtain sufficient data to show that A is not equal to B (i.e., the 
MA cross does not signal a price rise) and can say so at the 95 or 99% 
confidence level, I think I have disproven in a statistical sense the null 
hypothesis at some level of confidence PROVIDED my sampling was representative 
and the sampling size was sufficiently high to generate the proper amount of 
statistical confidence. I will have demonstrated my point with an error of 5% or 
1%. And of course, I must first establish that my dataset is normally 
distributed, otherwise my 'proof' is faulty. This analysis does not by any means 
constitute proof with certainty, which is impossible because one cannot sample 
the entire uinverse of data available. It only establishes a degree of 
confidence based on a sampling of a limited population of price data and the 
inherent error associated with it. Sample size is critical to any statistical 
analysis. The higher, the better. That's all statistics does. You may still act 
on the MA cross by buying, but you have been warned that if you do, the 
probability is very high that you will lose. There are all kinds of other things 
I haven't even touched on, such as sampling from a population of correlated 
stocks, or sampling only during a bull market, etc., etc. Nothing is certain in 
the game of prediction. 
 
Al Venosa
<BLOCKQUOTE 
>
  ----- Original Message ----- 
  <DIV 
  >From: 
  palsanand 
  
  To: <A title=amibroker@xxxxxxxxxxxxxxx 
  href="">amibroker@xxxxxxxxxxxxxxx 
  Sent: Sunday, October 26, 2003 4:55 
  PM
  Subject: [amibroker] Re: solving the 
  prediction problem
  Hi,I was speaking only of absolute proof and it 
  solely belongs in the realm of physical law or mathematical theorem, not 
  statistics.Regards,Pal--- In <A 
  href="">amibroker@xxxxxxxxxxxxxxx, "Al 
  Venosa" <advenosa@x...> wrote:> 
  Pal,> > Sorry, but I have to take the side of MarkF2 on this 
  one. Yes, I agree, you can prove anything you want with statistics, but 
  only if you use it improperly, which most people who have a vested 
  interest in something or are not objective in their analysis of an 
  empirical observation do (understand, I am not accusing you of anything; 
  I'm just stating a fact). If you use inferential, parametric statistics 
  on data that are not even close to being normally distributed, you can 
  prove anything. But your proof is flawed because you have misused a basic 
  underlying assumption that the particular statistical test was designed to 
  indicate. You speak of proof being only in the realm of physical law or 
  mathematical theorem. Perhaps true if you speak of absolute proof. 
  However, in statistics, all 'proofs' are based on measurement error, and 
  there is nothing on this planet that can be measured error-free (except 
  mathematical principles such as the area of a triangle or the Pythagorean 
  Theorem). In trading, the errors are huge, sometimes infinite, because 
  price behavior is so variable; so in order to try to 'prove' anything 
  statistically, you have to minimize that error as much as possible. How 
  does one minimize error when prices are determined by such a huge universe 
  of human beings, all with different outlooks on future behavior and all 
  biased to some degree by their own non-objective views? You can't. You can 
  only generate probabilities and confidence based on those probabilities, 
  and that is the heart of statistical analysis. So, statistics is not 
  meaningless at all and in fact is a very valuable tool in this 
  complicated trading business of ours. Use it properly, and you will be 
  rewarded. Use in unwisely, and you will lose big. > > Al 
  Venosa>   ----- Original Message ----- >   
  From: palsanand >   To: amibroker@xxxxxxxxxxxxxxx 
  >   Sent: Sunday, October 26, 2003 3:41 
  PM>   Subject: [amibroker] Re: solving the prediction 
  problem> > >   I do not dispute the usefulness 
  of statistics, only in the use of it >   to prove 
  something, which anybody who has a good grasp of statistics 
  >   in this planet knows, which apparently you don't.> 
  >   Regards,> >   
  Pal>   --- In amibroker@xxxxxxxxxxxxxxx, "MarkF2" 
  <feierstein@xxxx> wrote:>   > What planet are 
  you from?  Statistics *is* a branch of applied>   > 
  mathematics and, IMHO, the single most useful for trading.>   
  > >   > "Each of us has been doing statistics all his 
  life, in the sense >   that>   > each 
  of us has been busily reaching conclusions based on 
  empirical>   > observations ever since birth." -- 
  William Kruskal >   > >   > --- In 
  amibroker@xxxxxxxxxxxxxxx, "palsanand" <palsanand@xxxx> 
  >   wrote:>   > >   > 
  > Like I said earlier, Statistics is meaningless in the context 
  >   that >   > > you can come out with 
  statistics to disprove anything.  Proof >   > > 
  requires a law or a theorem.  Only Mathematics can prove 
  or>   > disprove >   > > and that 
  is one reason it is the greatest of all sciences.> > 
  >         Yahoo! Groups Sponsor 
  >               
  ADVERTISEMENT>              
  >        
  >        > > 
  >   Send BUG REPORTS to bugs@xxxx>   Send 
  SUGGESTIONS to suggest@xxxx>   
  ----------------------------------------->   Post 
  AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx 
  >   (Web page: <A 
  href="">http://groups.yahoo.com/group/amiquote/messages/)>   
  -------------------------------------------->   Check group 
  FAQ at: <A 
  href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
  > >   Your use of Yahoo! Groups is subject to the 
  Yahoo! Terms of Service. > > >   
  --->   Outgoing mail is certified Virus 
  Free.>   Checked by AVG anti-virus system (<A 
  href="">http://www.grisoft.com).>   
  Version: 6.0.525 / Virus Database: 322 - Release Date: 
  10/9/2003
  Send BUG REPORTS to bugs@xxxxxxxxxxxxxSend SUGGESTIONS to 
  suggest@xxxxxxxxxxxxx-----------------------------------------Post 
  AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: <A 
  href="">http://groups.yahoo.com/group/amiquote/messages/)--------------------------------------------Check 
  group FAQ at: <A 
  href="">http://groups.yahoo.com/group/amibroker/files/groupfaq.html 
  Your use of Yahoo! Groups is subject to the <A 
  href="">Yahoo! Terms of Service. 
  
<BLOCKQUOTE 
><FONT 
  face="Courier New">---Outgoing mail is certified Virus 
  Free.Checked by AVG anti-virus system (<A 
  href="">http://www.grisoft.com).Version: 6.0.525 
  / Virus Database: 322 - Release Date: 
10/9/2003






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.