Intensive Review Genetic Builder Intensive Review

[jamesbeale]

I have only recently come across forexpeacearmy.com, and I'm glad I did.

This review of StrategyQuant Genetic Builder is very good as far as it goes, and timely for me. I've known about StrategyQuant genetic builder for some time, but have hesitated to buy into it, and felt that a 2 week trial period was really just a waste of my time.

What puts me off it was the very thing to which Pharaoh referred early on as a strength. Even with the "coding power of a few hundred thousand code monkeys that can all type very fast" I really wondered if the production of truly worthwhile EAs was possible in any meaningful time, without the intervention of luck.

A little searching on Google and you'll discover this website: Infinite monkey theorem - Wikipedia, the free encyclopedia

Scroll down the page a bit, and under "Random document generation" you'll see the following:

One computer program... had worked for (the equivalent of) 42,162,500,000 billion billion monkey-years (when) one of the "monkeys" typed, "VALENTINE. Cease toIdor:eFLP0FRjWK78aXzVOwm)-‘;8.t" The first 19 letters of this sequence can be found in "The Two Gentlemen of Verona". Other teams have reproduced 18 characters from "Timon of Athens", 17 from "Troilus and Cressida", and 16 from "Richard II".

The universe isn't that old. Indeed, it will take the power of a new generation of quantum computers to be able to process something like this, even then... who knows how long? I don't!

This is not encouraging. It doesn't leave much hope for me with my silicon variant, even if it is suped-up and running extra fast Linux. But, of course, I assume I'd need Windows for Genetic Builder...

Mark Fric (StragegyQuant author) will counter that Genetic Builder is using considerably fewer variables. But there is a very large number of possible combinations. Again, Mark will no doubt say, but we are focussing down to certain knowns, and the Genetic Builder builds on what it finds to be good, thereby being somewhat Darwinian in its approach. But again, if we do focus down on some knowns, we may miss out on that truly spectacular left-of-field EA that no-one could have predicted. And a Darwinian approach can also end up in a cul-de-sac, such as the dinosaurs were. It took a meteor to budge them, to let us get a look in.

To my mind, the impression I have from reading Pharoah's (and others) excellent work here is that they haven't stumbled upon that amazing EA, or even a series of pretty good ones all adding up to amazing. Indeed, if Mark Fric had, I think he would have closed StrategyQuant down by now, saved himself the bother of dealing with users and be cruising the Mediterranean, or in some other way be a non-dom tax exile on a mega-yacht. Or if he is already, then he is truly altruistic to be dealing with the likes of me!

However, before the reader goes off thinking I'm completely blinkered and sarcastic, I think that Genetic Builder is an excellent concept. I also believe, mostly from my reading here, that it is well put together and supported by Mark. All credit to him. I'm also not saying I won't at some stage buy into it. I might. But there again I'd have as much chance of winning the National Lottery this year, sometime, with the money not spend on Genetic Builder as finding that elusive EA.

But please do - someone! - tell me I'm wrong. I'd like to think I am.

I'd like to finish by saying three things:

Do your homework. There is no better way of finding out what others have found to work. Look back in history.

Learn logic. If you are not a programmer (and few of us are really gifted with this), then use a builder, and create your own EAs based on what your have learned from the work of others. We all stand on the shoulders of giants.

Money management is key to long term profits. I believe most traders completely miss this point.​

Most important of all, have fun!

 

[Batchboy]

Hi James,
Even Mark will say there is no holy-grail, but I like the term "near holy grails". Anotherwords, acceptable profit in MM mode, acceptable dd throughout, acceptable monte-carlo dd, acceptable #trades each month. Currently found one EA that avails eurusd 1K to 170K in 11yrs on 2% MM w/ 20-26% dd. What makes these dynamic is lack of tick sensitivity using bar-open-only mode in its tests and tests on mt4. I'm ok w/ 4 trades on average per month (typical results for SQ) but a forex EA tester site owner is a lot tougher, so it may just endup in my hands, not sure yet. There are enough variables in the formula so over 11yrs 15minTF the optimization run is 2 weeks so far and will be continuing at least a few more days.
I'm exprimenting with an idea to produce signals, that of replacing with newly created EA's when each one's performance is doubtful. So far am up a few hundred pips. I use ibfx live mini acct.
I'm intrigued by thought of using indy dependent indies so that such indies depend on the market. I did find one dynamic solution on 2yrs data, but have yet to try on many years due to momentary technical issues.
It does no good to play casino guesser as to how well one can do with such a massive software, one can only continue like Edison, try, try & try. If your not like him and presume to evaluate before using intensively for a year (practically wearing a lab coat) then this isn't for you.
Thx for listening.
Jerry
PS: Mark will let yoh have trial for longer than 2 weeks if you think you need it.

 

[Pharaoh]

My available free time has been almost nil for quite a while now. If/when I get some time, I'm setting up a second computer just to run the newest version.

From what I can see, the latest upgrades take it from a million monkeys with a spell checker to a million monkeys with spell and grammer checking. Adding in the mix and match mutations from the best performing random ones, and I believe there is real potential to create some useful EAs.

Like all EAs, they wouldn't be eternal grails - the market itself changes over time. A perfect EA in 2012 may not be so perfect in 2013 without considerable retuning. In some cases the market changes may be sufficient that a previously good EA may become irreparably useless. Of course, that's why one needs to keep those monkeys busy generating new code.

 

[jamesbeale]

Dear Jerry and Pharaoh

I'm very interested to read what you say.

I think I may have obfuscated my thoughts somewhat. Yes, Genetic Builder is a good idea, and one possibly to consider placing in one's armoury of trading tools. But that's the point; its just one, and probably not the most important one. To any reader coming afresh to this thread, perhaps as I did whilst researching, and wishing to discover ways of creating and testing new and novel trading approaches, I'd like to say this; there is no substitute for homework. Study the charts and get really familiar with how the market works. Read the works of others. I found W D Gann's "New Stock Trend Detector" enlightening, and he was published in the 1930's.

This is also where I slightly diverge with Pharaoh. Yes, the market does evolve over time, but usually it reverts too. This is due to the nature of the human race, a factor none of us in meaningful numbers can change - and this is a good thing for traders. Indeed the successful trader is one who can subvert his own human tendencies somewhat. So if an EA worked last year but not this, that's probably just because the trading signals it depends on were less prevalent in the current year. Therefore have more than one method of trading the same market to achieve diversification. And do test your EAs against data for at least ten years back.

Genetic Builder may be a good idea, but to readers of this thread in general I say, don't rely exclusively on it to produce EAs. The chances of lifting yourself to the next level by this route alone are against. Do your homework too.

Cheers,
James.

 

[Pharaoh]

i think of SQ/GB as a different approach. Building and tweaking an EA from one's own trading ideas is a very good thing to try. Yes, many things have been tried before, but human ingenuity can always come up with new ones, and some may end up working very well.

On the other hand, SQ/GB takes a totally different approach. From a set of initial parameters, it tries more things in a day than a coder could try in a year. The ability to filter out the useless results is what gives it the power to have its own potential to build something that may work well.

The good news is that it's a big market, so there's plenty of opportunities available for both approaches.

Now if only Mark will combine a DIY EA builder with the broad array of testing, mixing, and mutating concepts that are in StategyQuant.

 

[Batchboy]

Hi folks,
Came across a stupendous strategy that takes 1K to 760K in the 12yrs that had passed using 2.5% MM.. Its with 27% dd and only 6% stagnation period. Its on 15minTF eurusd. It has "only" 2 trades per month. Do you think the market wouldn't care for it since its kinda boring waiting for trades? Or do you think a percentage can like such?

It sounds like you bought the unit Pharoah, can't you just leave it running with computer staying on? That be better than just being here philosophizing about SQ, as like me, you never know what you may come across. Just arrange computer and organize life so that you don't have to bother with it much, just a rare checking in to see how "she" is accumulating "monkeys", lol..

My big disappointment is, to make a portfolio (which SQ can stick together and evaluate) I'd need similarly performing strategies to go with this hot one, and since this eurusd strategy is such a rare find I don't feel confident. Well, **maybe** there is hope with Geneticizer and Improver sections.....maybe, I don't know.

Mark has always said the std indies would do it, so far my custom indy tries hadn't gotten anywhere and this hot eurusd is completely with std indies, so thus far he is correct.

But for 12yr searches, immediately going to other pair like gbpusd shows great stumbling block in getting anywhere, databank accumulation of even the weakest puny profiting strategies is extremely slow on even an i7 dual core.

Ok, just passing thru so thought to squawk.

Jerry