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

Re: [amibroker] Re: A Puzzell: Otimize winning %



PureBytes Links

Trading Reference Links

Dear Dimitris,

Customizable columns in optimizer are considered.
As you guess, the most flexible solution would be to
have an ability to write the formula for calculation
of each column. This unfortunatelly adds additional complexity 
to the program. Therefore I plan to address this issue
in some more universal manner which is a part
of my "secret plan" of making AmiBroker the most customizable
T/A platform ever. This, however, has to wait until
my Q4/2001 plans are implemented (releases 3.8-3.9). 

Best regards,
Tomasz Janeczko
amibroker.com

----- Original Message ----- 
From: "DIMITRIS TSOKAKIS" <TSOKAKIS@xxxx>
To: <amibroker@xxxxxxxxxxxxxxx>
Sent: 25 September, 2001 10:49
Subject: [amibroker] Re: A Puzzell: Otimize winning %


Dear Tomasz,
For many applications it would be wonderful
to optimize and have user defined columns.
(not exclusively related to buy-sell conditions
and profits results)
Example:I try to estimate total stochd O/S time
for various O/S levels (say 35 to 10, step 5) and 
find the proper level that gives 10% total O/S time.
In other words, mathematically speaking, to be able
to solve through optimization, first degree equations
in Amibroker world.
(For now we mostly concern with identities)
Will it be possible for the future ?
If someone wants to find the sqrt(2), there are some
artificial techniques, but there is a simple and frienly
procedure:Since he knows that 1<sqtr(2)<2, he may
try all numbers from 1 to 2, step 0.1 or any other and
get sqrt(2) at desirable accuracy.
Archimedes, at 200 B. C. invented this method and 
clearly understood that there is no reason for the
"exact"-mania of Euclidean predecessors.
For him the relation
3 1/7<pi<3 10/71 
was a perfect way to know and use pi and then apply
exaustion method to increase accuracy at any level.
Ancient Greeks stopped evolution because of "asymmetry".
Newton and Leibnitz received Archimedes message with 
a 17 centuries delay and reopened this wide door to "homo sapiens".
Since with optimization we follow the same old Archimedean path,
is it possible to extend its use beyond profit results?
It will change the way we think in Amibroker Logic.
Best Regards
Dimitris Tsokakis 
--- In amibroker@xxxx, "Tomasz Janeczko" <amibroker@xxxx> wrote:
> Hello,
> 
> The solution is:
> 1. Run the optimization
> 2. Export result list to CSV file.
> 3. Load it to Excell
> 4. Create a new column with # of winners divided by #of loosers
> 5. Sort by this new column.
> 
> Best regards,
> Tomasz Janeczko
> ===============
> AmiBroker - the comprehensive share manager.
> http://www.amibroker.com
> 
> 
> ----- Original Message ----- 
> 
> From: <b519b@xxxx>
> To: <amibroker@xxxx>
> Sent: Tuesday, September 25, 2001 12:42 AM
> Subject: [amibroker] A Puzzell: Otimize winning %
> 
> 
> > Is there a way to get the optimizer to optimize for the 
percentage of 
> > winners rather than the profit percentage? The Report feature 
gives 
> > such a percentage but the Report feature is only available after 
> > backtesting, not after optimization.
> > 
> > I am working on a market timing indicator for 
identifing/predicting 
> > changes in the market trend and at this stage of my work, I am 
more 
> > interested in getting a high win percentage for the indicator. My 
> > plan is to use the indicator to identify/predict changes in 
market 
> > trend, and then to use a different set of selection criteria to 
pick 
> > the stocks to buy or short.
> > 
> > Right now I eyeball the results by having the optimziation 
results 
> > sorted by number of winners, but this is inefficient since the 
total 
> > number of calls varies considerably. Getting the results ranked 
by 
> > winning% would be much more easier for me.
> > 
> > b
> > 
> > 
> > 
> > 
> > 
> > 
> > Your use of Yahoo! Groups is subject to 
http://docs.yahoo.com/info/terms/ 
> > 
> > 
> >





Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/