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

RE: [amibroker] Re: Real-world trading (specifics)



PureBytes Links

Trading Reference Links




<SPAN 
class=270563220-13042003>Thanks, Fred Tonetti, for the comments.  I 
will endeavour to answer your questions without "giving away the 
farm".
<SPAN 
class=270563220-13042003> 
First, 
I have to tell you that I think my approach is different from MOST of the people 
in this group.  Of course, that could mean that most of 
the others have it right!  
<SPAN 
class=270563220-13042003> 
I 
would NEVER optimize for a different set of parameters for each 
stock.   I'm of the opinion that MSFT can start looking like IBM 
tomorrow.  AOL can take on the look and feel of INTC next 
week.   So, I want to have one set of parameters that works on ALL 
stocks over at least six years of data (preferably ten to twelve).   I 
end up with a set of parameters that works over 13,000+ stocks (active and 
extinct) times however many days those stocks have traded.   The 
number of trades can be between 15,000 and 50,000 and gives me some feeling that 
the system(s) will be robust in the future.
<SPAN 
class=270563220-13042003> 
The 
next area I seem to treat differently than most as well.   I trade 
fixed size positions all the time.  My backtesting and realtime 
trading is always based on fixed position size.   If I have cash, I 
will take as many trades as I can take.  If I don't have enough cash to 
take every trade, I will sort the orders by "something".   If I'm not 
getting enough signals to use all my cash, I will gradually increase the 
bet size.   AB lends itself to this approach, although I would like to 
see it more automated.  I'll give an example:
<SPAN 
class=270563220-13042003> 
Let's 
say I'm sitting on $100,000 cash, or will be after I close out some positions 
tomorrow.   If I'm trading $10,000 per transaction, obviously I have 
enough cash to take ten new positions.   My system may generate 100 
orders for tomorrow.   I will add a column to my exploration so that I 
can sort by it (or at least look at it).   For simplicity, let's say 
that I know that my system works better on low price stocks; the lower the 
better.   That's almost too easy, but I could sort my buy orders by 
closing price and take the first ten.   Obviously, I would have had to 
backtest this premise before trading in realtime.   

<SPAN 
class=270563220-13042003> 
I sort 
by whatever I have found (via backtesting) gives me the best 
results.   Since I have other information in my data (fundamentals, 
etc.), I can sort by just about anything you can imagine.   

<SPAN 
class=270563220-13042003> 
You 
quoted me as saying that I use volume * price as one of my filters.   
I use something between a ten-day and fifty-day average volume times the 
(actual) closing price to give me the turnover.   If I'm trading 
$10,000, I want the turnover to be at least $200,000.   That's not a 
science, just a judgemental ratio of my order size to the average turnover in 
order to get an easy fill (in and out).
<SPAN 
class=270563220-13042003> 
I hope 
I have answered your questions.  If not, hit me 
again.  
<BLOCKQUOTE 
>
  <FONT face="Times New Roman" 
  size=2>-----Original Message-----From: Fred Tonetti 
  [mailto:ftonetti@xxxxxxxxxxxxx]Sent: Sunday, April 13, 2003 4:22 
  PMTo: amibroker@xxxxxxxxxxxxxxxSubject: [amibroker] Re: 
  Real-world trading
  <FONT face=Arial 
size=2>Chuck,
  I can tell from 
  messages that you've post<SPAN 
  lang=en-us>e<SPAN 
  lang=en-us>d in the past that 
  you are thorough in your testing ...
  Without giving away the 
  farm as it were I am interested in the kinds of systems you 
  develop<FONT face=Arial 
  size=2> and trade<FONT 
  face=Arial size=2>, do they in general look for certain kinds of conditions 
  like for example a Raschke "grail" set up or do they more belong to the timing 
  of things with in the larger trend or ?
  I think you are far 
  beyond most here or at least me in terms of how you go about selecting stocks 
  to trade beyond whether or not they match some pattern in a generalized 
  way.  I've heard you speak of volume > X etc., but I think most 
  here<FONT face=Arial 
  size=2> or at least I 
  would benefit from any and all information you'd be 
  willing to share about how you pare down 10,000 tradable issues into something 
  more manag<FONT 
  face=Arial size=2>e<SPAN 
  lang=en-us>able on a real<SPAN 
  lang=en-us><SPAN 
  lang=en-us> time 
  basis.
  Regards, 
  FredSend 
  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. 







Yahoo! Groups Sponsor












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.