Forex Strategy Builder and Trader

[slipshod]

Doug, huge thanks for bringing this to everyone's attention, I'm really enjoying prototyping ideas in this program

A question for Miroslav, or anyone else who might know, is there any way to compare chart-based indicator values (as distinct from oscillators), in particular MAs and the various types of bands? I know there's an MA Cross, but often more than that's required. If I wanted to write a filter that says "only enter the trade if this MA is above that MA" how would I do it? Or "enter if the low of this band is above the low (or high) of this other band". Or "enter if the low of this band hasn't been below this MA in the last X candles".

Also the Generator seems to use many indicators/oscillators that aren't on the list to disable, thus making the generation take a lot longer than it perhaps needs to.

Oh and lastly, unless I'm missing something there doesn't seem a way to make the Generator build on an existing strategy, rather it always wants to create new ones regardless of whether the "Generate a new strategy at every start" checkbox is ticked or not.

cheers & thanks in advance,
Andrew

[Popov]

is there any way to compare chart-based indicator values, in particular MAs and the various types of bands?

Each indicator works independently. There is no way to compare two indicators using only the visual interface. To do so, you have to write a custom one. You can call and initialize indicators and to write the specific logic. Here is an example of "Momentum MA Oscillator". It creates and sets a Momentum and a MA, and compare them:

Code: Select all

var momentum1 = new Momentum();
momentum1.Initialize(SlotType);
momentum1.IndParam.ListParam[1].Index = IndParam.ListParam[1].Index;
momentum1.IndParam.ListParam[2].Index = IndParam.ListParam[3].Index;
momentum1.IndParam.NumParam[0].Value = IndParam.NumParam[0].Value;
momentum1.IndParam.CheckParam[0].Checked = IndParam.CheckParam[0].Checked;
momentum1.Calculate(DataSet);

double[] adIndicator1 = momentum1.Component[0].Value;
double[] adIndicator2 = MovingAverage(period2, 0, maSignalMAMethod, adIndicator1);

for (int iBar = firstBar; iBar < Bars; iBar++)
{
    adOscillator[iBar] = adIndicator1[iBar] - adIndicator2[iBar];
}

The full code is here: Momentum MA Oscillator

Also the Generator seems to use many indicators/oscillators that aren't on the list to disable, thus making the generation take a lot longer than it perhaps needs to.

Really? It should be a bug. Can you show a screenshot of an indicator that is banned and latter appeared in the generated strategy? You know that you have to ban an indicator for the specific slot type, right? I mean that if you ban an indicator for "Opening logic condition" only, it can appear in "Closing logic condition" slot.

there doesn't seem a way to make the Generator build on an existing strategy, rather it always wants to create new ones regardless of whether the "Generate a new strategy at every start" checkbox is ticked or not.

That's correct. If you do not lock or link some slots, Generator can delete or overwrite them.
"Generate a new strategy at every start" sets if Generator starts selecting strategies from zero profit or form the profit already reached by the loaded strategy.

Generator is stupid. It has no a conception of a strategy as a structure or general idea. It selects random indicators, selects random params, put them all together and checks if the resulted backtest satisfies the criteria and if gives enough profit.

[slipshod]

Hi Miroslav, thanks for your detailed reply

is there any way to compare chart-based indicator values, in particular MAs and the various types of bands?

Each indicator works independently. There is no way to compare two indicators using only the visual interface. To do so, you have to write a custom one. You can call and initialize indicators and to write the specific logic.

That's what I thought might be the case - no problem, will start coding

Also the Generator seems to use many indicators/oscillators that aren't on the list to disable, thus making the generation take a lot longer than it perhaps needs to.

Really? It should be a bug. Can you show a screenshot of an indicator that is banned and latter appeared in the generated strategy? You know that you have to ban an indicator for the specific slot type, right? I mean that if you ban an indicator for "Opening logic condition" only, it can appear in "Closing logic condition" slot.

Nevermind, newbie mistake on my part (won't be the first or last!)

there doesn't seem a way to make the Generator build on an existing strategy, rather it always wants to create new ones regardless of whether the "Generate a new strategy at every start" checkbox is ticked or not.

That's correct. If you do not lock or link some slots, Generator can delete or overwrite them.

Ah, the word "lock" answers my question, thanks!

[Popov]

There is a dropdown list box just over the indicator list. On your screen shot, it's set to "Opening Point of the Position". The list doesn't include RSI, because RSI is not an indicator that can be set to Opening Point. If you change the selection of the list box to "Opening Logic Condition", you'll see RSI.

[slipshod]

Yep, realised that shortly after my post. Sorry for the trouble, I tend to think out loud on a forum when I get stuck with something, a consequence of trying to do too much too quickly in too little time

FSB is simply amazing btw. When I think of all the hours I've wasted trying to wrangle meaningful results out of Empty4...

[slowkey]

Yep, realised that shortly after my post. Sorry for the trouble, I tend to think out loud on a forum when I get stuck with something, a consequence of trying to do too much too quickly in too little time

FSB is simply amazing btw. When I think of all the hours I've wasted trying to wrangle meaningful results out of Empty4...

No probem Slipshod we all get to learn this way. I have read studies that people who think out loud or in my case talk out loud have orderly minds . So the steps you took to come by the answers you seek are to every members advantage. Thanks for the kind comments. Thanks Mr Popov for being so generous with your time here to answer questions.

Doug

[slowkey]

Doug, huge thanks for bringing this to everyone's attention, I'm really enjoying prototyping ideas in this program

Andrew

That by the way is a great way to think about the generator. It is a very nice prototyping tool. It is not designed to pop out ready made strategies that will make you money on first blush. There is no Holly Grail software like that yet. It is a very nice tool to hammer out ideas that you might have or it may give you insight about. That moment to do something special.

You can also work through the logic console on the main screen in the middle and do the same thing with out going into the generator if that is a more comfortable approach or more your style.



Doug

[slipshod]

Yes I certainly wouldn't recommend just turning on the Generator for a 12 hour run, walk away then come back hoping to find an optimized & robust trading system waiting for you!

So far I'm finding the most value in exploring ideas manually in the main screen, then using the Generator to look for filters and/or better exit parameters.

Two things are essential I believe:-

1. test on large amounts of historic data. Back to early 2005 at least.
2. test strategies on multiple currency pairs. If it produces a lovely rising curve on one but margin calls on everything else, chances are its over-optimized for one currency. It'd be nice to be able to select multiple pairs for the Generator as well ... can't have everything I guess

[slowkey]

Yes I certainly wouldn't recommend just turning on the Generator for a 12 hour run, walk away then come back hoping to find an optimized & robust trading system waiting for you!

So far I'm finding the most value in exploring ideas manually in the main screen, then using the Generator to look for filters and/or better exit parameters.

Two things are essential I believe:-

1. test on large amounts of historic data. Back to early 2005 at least.
2. test strategies on multiple currency pairs. If it produces a lovely rising curve on one but margin calls on everything else, chances are its over-optimized for one currency. It'd be nice to be able to select multiple pairs for the Generator as well ... can't have everything I guess

I actually have been thinking for a while now that we may never find the perfect ea that will run on all major pairs. I am not sure you call a strategy that runs really good on one major pair over-optimized because it can not run on multiple pairs. All the pairs are so different and have such different volatility and personalities you might say that optimizing for one pair may not be a bad idea for success. If I can get my usdjpy strategy to run very consistently on the 1 hour time period that would make me very happy. Fine tuning it from time to time as the market changes. A program like this lets you take that approach to automation. I just have my doubts about an EA that can become a holy grail EA on several different pairs. I am not against trying but I just think it easier to fine tune to one pair and I do not think that so much over optimized to one pair as it is a good fit to one pair if it consistently runs well accept for a few minor tune ups from time to time.
Just my 2 cents
Doug

[slowkey]

Over optimized, I have no idea but it is going on demo. However sort of an ideal real life equity curve.

Perfect equiity curve.jpg