Forex Software — CURVE FITTING

Re: CURVE FITTING

2015-02-07T19:51:39Z

For the FSb we need a
1- Walk-forward testing it will validate the strategy very well
2- visual comparison of the optimization results for the main strategy '' because in my opinion'' selecting an optimized strategy needs human intervention.

For my opinions:
3- Reduce the number of parameters system has, the more parameters, the more opportunities of curve fitting

4- Not optimizing the OOS segment otherwise it will loss the concept behind it so only optmize the IS segment

5- finding the stable segment of optimized results to select from '' if the highest result is considered as an outlier so most probably it will be wrong '' we need point No.2 to do the comparison''.

Re: CURVE FITTING

2015-02-07T17:45:49Z

I think that the next challenge for FSB Pro is to trade the news releases.
Their impact is enough to drive to big profit if someone could interpret them
flawlessly.
I have some thoughts about price patterns and candlestick patterns but
I think that they must be combined with news releases interpretion
and this will help us to generate consistently profitable strategies
which will have sense and combιne techical analysis with fundamental analysis.

I think that Mr Popov will find a way to integrate news releases in FSB Pro.
Integration could be done with a seperate file for example which would be the same
for all trading pairs in FOREX. Personally I use Zulu trade calendar to keep in touch.
Each kind of news could have a code like coordinates:
Country - Importance (high medium low) -Kind ( Interest Rate Decision - Employment etc.
Also it would be useful for someone to adjust it to his brokers time region.

Re: CURVE FITTING

2015-01-18T14:33:22Z

Unfortunately all the mathematical tools are two edged.
This is the reason that the traders must have sense
like all the other experts to their sector like music, maths etc.
For example in real life we use almost always curve fitting with excellent results.
For exampe the Winter Summer Winter Summer series in a specific place on Earth
has the same distance in days year by year during our lifetime with minor differences.
By this way all the spieces on Eath can be appropriately prepared for Summer and Winter.
Of course it could vary thousands years earlier or later but the point is the same.
Every one who has the ability to feel the seasons will absolutely be sure about
Winter and Summer.
For example some days in Winter in my city are like Summer and vice versa.
BUT THE PLANTS AND THE ANIMALS do know it and they are not outsmart.
I placed a physical example because the nature is a great teacher for us.

So I believe that the point is the Correct Curve Fitting and not Curve Fitting.

I will agree with Blaiserboy about his suggestions because his suggestions
mean in other words that somebody must be a trader in order to make winning
strategies.
First somebody must learn a little about trading and then he will be able to generate
some strategies and test his assumptions.

Every one of us must define his Summer and Winter and try to find them
discarding some Sunny Winter Days or some Rainy Summer Days from
his research.

For example the plants don't oursmart about some Sunny Days in Winter
because the Day Duration is smaller than Summer etc.

By this way we must add some secure conditions to our strategies
and this will drive us to the correct path.

Finally I close with a suggestion.
There are Signal Providing Services.
Every one who has winning strategies and he feels that they are interesting
can become a signal provider and by this way he will help other people to earn
some money too.
This for FSB Pro means that all the people must try FSB Pro and learn how to
use it, make winning strategies and post their ideas.
Perhaps they will be able in the future to earn some money
helping other people to profit from FOREX via Signal Providing Services as
Signal Providers. Also they can make some money as followers too.

CURVE FITTING

2014-12-02T16:29:16Z

Of course, using the generator to develop a strategy will result in curve fitting, and the longer we run the generator to develop the strategy...... of course, the more curve fitting.

Some Remedies

Do a walk forward using the dates tool in the History Center

Use standard indicator values

Use short generator times, as the longer the generator runs - then the better the fit

Monte Carlo

Multimarkets

Comparator

I think that these should be able to reduce curve fitting significantly.

I welcome suggestions as to other means to reduce curve fitting. Please add your ideas to this thread.

If there is some technique that can be added to FSBPro, please describe it here so that Popov can consider adding it to the software.

As there will be some degree of curve fitting in every strategy, we have to accept that it exists.

Also, because each strategy is going to have some type of difficulty we are going to need more than one strategy.

Because each strategy is going to have periods that it will not be trading, we will have to have more than one strategy.

The portfolio feature in the program will be able to demonstrate what is needed in order to fill in the gaps.

Please add whatever ideas you may have to this thread regarding management of Curve Fitting.

Thank you

Good Trading..!