InstaForex Contest

[fapturbo]

Here's one for you Norm... Give your EA a test... Enter it in this contest and win some pocket change...

http://instaforex.com/forex-contest-table.php?lang=en

 

[Stormin_Norman]

theyre not going to 'steal' the EA though?

 

[electronicmaster]

theyre not going to 'steal' the EA though?

I see a lot of Russians on the list. Be careful.

 

[Stormin_Norman]

trust noone

the 1st prize is only $8k USD. not worth the risk i dont think.

 

[BentRod]

theyre not going to 'steal' the EA though?

So that is why these comps are run.

How do they go about stealing the actual EA anyway, doesn't it reside on the client side only?

 

[Stormin_Norman]

So that is why these comps are run.

How do they go about stealing the actual EA anyway, doesn't it reside on the client side only?

im not IT expert, but i do wonder what they could have hiding in the back end of their trader or something.

am i being paranoid? if i can be convinced i am ill enter it and run it on high leverage and see how i can go; otherwise there are bigger fish to fry.

 

[Cartman]

im not IT expert, but i do wonder what they could have hiding in the back end of their trader or something.

am i being paranoid? if i can be convinced i am ill enter it and run it on high leverage and see how i can go; otherwise there are bigger fish to fry.

yes u r being paranoid Norm lol ---- but u know what they say ----

Just because u r paranoid ---- it doesn't mean people aren't out to get you

(just kidding ----- good onya with the Bunyip btw --- )

 

[wabbit]

Send me all your EAs for a competition; oh, and by the way, your EA cannot call other .dlls which might hide the smarts and logic of your system. All the workings must be contained within the MT4 compiled EX4 file (which is easily decompiled).

So which EAs deserve to be decompiled? The ones observed to perform the best of the month of the competition! i.e. the competition winners.

The no .dll rule hasn't been specified for this competition (or at least not that I could find easily) but they do have the ubiquitous statement in their rules:

"Registering in the Contest, participant accept all the restrictions and rules of the Contest."

If there are other rules involved, why aren't they listed on the rules page?

The MQL ATC had a prize pool of USD$80K (cash prizes) which is not to be sneezed at. Another competition has a prize pool of USD$8K (plus some trading vouchers) IMHO, eight thousand dollars is buying a serious amount of EAs given the registration interest so far.

Fishy? Maybe? Maybe not?

I am considering putting up a shot though. I probably wont win but then again, if I do, happy days. Gotta be in it to win it. I might up the extremely high risk, goes bust at random times martingale system and see how it goes? With 100K paper money, it might just survive the month?


wabbit

 

[Stormin_Norman]

you convinced me to sit it out wabbit

 

[Cartman]

if i stayed up all night every night would this stand a chance with some tweaking ?? --- last nights + a few scalps today ---


ps only demo on a new system so not getting excited ---- but the ducks r lining up for another crack thats for sure

only 1 or half lot trades as well ---- no biggies ---

 

[wabbit]

One of the things with these competitions is that real money management and trade management goes out the window in most cases. You can trade in the comp like you'd never do in real life. The whole idea is to make as much money in the period of the comp without any actual risk. See the results of the last ATC to see how many EAs bet the house on the first trades (and their rules were no more than three open trades at one time, maximum lots of 5 per trade which defeated the martingale gamblers, no scalping either).

One of the guys who finished in the top ten of the ATC didn't trade for the best part of the championship. His EA triggered a "chicken switch" which meant that it wouldn't take any more trades. For two months his EA sat there doing nothing; other systems overtook it, then eventually the market changed creating some massive losses and the dormant EA wound up doing better than a lot of other systems.

Just tossing around an idea: does anyone want to form a syndicate to enter in the comp? Or is everyone happy doing their own thing?


wabbit

 

[sinner]

Hi wabbit,

I've never written an EA or anything really, but lots of coding experience. If you are looking for a geek for your syndicate consider me

 

[wabbit]

Hi Sinner,

Unless the concepts get extremely complicated, I don't think there would be much coding to do??

The hard part to these competitions is find the right ideas that can make huge profits within the rules of the competition and its time. Sometimes, the humble MA crossover might do ok, sometimes a stochastic system will do well... I think to beat the competition requires the right combination of extreme risk taking with a reasonable entry criteria.

Again, its the ideas not the code that pose the problems.


wabbit

 

[pilbara]

His EA triggered a "chicken switch" which meant that it wouldn't take any more trades.

hey that's a good idea, kinda like training wheels for the EA. I might try that on demo, giving the EA a maximum drawdown where it must exit, and a maximum profit where it should exit. So the EA will aim for a weekly profit target, or hit my maximum drawdown limit, whichever happens first.

 

[Punisher]

I see a lot of Russians on the list. Be careful.

Yeh nah.
Have you had a bad experience EM?:bazooka:

 

[peter2]

How about you code up Stormin Norman's "stochastic entry with the dominant trend" strategy?

Crank up the risk to the comp limits and let 'er rip.

 

[Stormin_Norman]

as long as i get my share of the winnings that said the system i use i am sure is used by other traders, it is too simple and straight forward not to be. i cannot claim to have invented moving averages and stochastics; nor the combination of their use. i only saw the relationship between moving averages and stochastics working together to indicate price movement - as im sure others have.

that's the code im currently working on. that system requires more human reading then i first realised; and getting a computer to think through that is quite a challenge. that said each addition/revision of the code is proving successful.

im currently stressing out my testing computer with tests for version 3. attached is the version 2 - which is the system i posted, with some additional methods/indicators added to help the computer take into account factors which a good, experienced trader does almost as 2nd nature. a big addition still needed to the automated system is on exits. current work has focused on entries which are the easiest part of a trading system - entries are far harder and more neglected. the system so far has worked on constant values for stop loss, move to break even and take profit amounts. the work on a variable, flexible close out of positions (be them at a profit or loss) is the key to this system.

this is the result of a randomly selected market (AUDUSD) over 4 years. i suspect any market (accounting for each's personal quirks) will respond in generally the same manner.

 

[wabbit]

that's the code im currently working on. that system requires more human reading then i first realised; and getting a computer to think through that is quite a challenge. that said each addition/revision of the code is proving successful.

This is something I encounter VERY frequently when clients approach me to code their systems. Their eyes are full of profits as all they can see in their charts is how their system will make them money. What they don't see is the opportunities for their systems to lose money, or the interpretation/interpolation they do when they think they are trading "true" to their system. The human brain and its visual sensors are amazing products. Nothing compares to their ability to analyse and see patterns/scenarios which computers cannot; computer are too literal and even when you implement "fuzzy logic" the boundaries are still not broad enough. The human uses terms like "just about to", "comes close to", "is approximately equal" etc when computers don't know how to handle these scenarios. Think about Fibonacci numbers: how often is the retracement exactly 0.618? what if it's 0.619 or 0.617? What about 0.7 or 0.6? How close is "close"; how is "distant" quantified in a computer program?

This is why many people are very disappointed when they see the results of their system which don't work when the computer is left in charge, and the human brain and all its interpretative powers are removed from the processing.


wabbit

 

[electronicmaster]

Yeh nah.
Have you had a bad experience EM?:bazooka:

Trust me. I have worked with a Russian, and he was just like Borat. Yea na manda hren :swear::chainsaw:

EDIT:- No No No, he is from Kazakhstan, but clams to be Russian.

 

[pilbara]

Think about Fibonacci numbers: how often is the retracement exactly 0.618? what if it's 0.619 or 0.617? What about 0.7 or 0.6? How close is "close"; how is "distant" quantified in a computer program?

to give an example, yesterday I was looking at something where I thought a parameter should be around 0.5, but I thought the fibonacci 0.618 might be better, so I tried 1/2, 3/5, 2/3 and 3/4. As expected 2/3 and 3/4 were worse than 1/2, but 3/5 which is roughly 0.618 was significantly better. I just left it as 0.6, because it might just be an upper bound on 0.5, like within a tenth of 0.5.

Regarding "close" and "distant", I think you need to use a tolerance when comparing, and this tolerance should be derived from the current level of volatility. If the price is moving fast, a wider tolerance is needed.