Back test starting from $100 to over 180 million | Page 4 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Back Test vs Forward Test 6 replies

High Leverage Trading, Going For A 100 Million, 100% A Week 149 replies

High Leverage Trading, 100k To 100 Million, 100% A Week 23 replies

Why EA test is forward test better than back test?? 1 reply

Cash & Carry (Grid) took over 180 pips out of the market yesterday and under 1% DD 1 reply

Attachments: Back test starting from $100 to over 180 million
Exit Attachments

Back test starting from $100 to over 180 million

  • #61
  • May 16, 2015 1:37pm May 16, 2015 1:37pm
Quoting CuriosityFX
You can not randomize your way to a profitable 'EA'. You can not stumble upon a profitable set of parameters and indicators for an 'EA'. You can not trade with "Expert Advisors" that are somehow experts without you yourself being an expert. The only profitable expert advisors you will find were 'profitable' in backtests due to chance, not due to an understanding of the market. Once you become an expert, you don't use "Expert Advisors", you use algorithms or models. The algorithms don't come from stumbling around, they come from years of research...
This is pure bullshit.

Not a single shred of truth to much of any of it, really.

Thinking that you need to use anything but the tools that are provided to be profitable in market, lends itself to somebody who has only been able to fail at market both manually and through the use of Automated Trading.

Each EA, successful or not, has been written by an understanding of market by the use of specific optics and a lot by guiding rules of technical systems. Application of any metrics of market through automation can provide definitive information that can take thousands of hours of watching to realize. This is the base of stumbling around with a limited time commitment.

How do you define a trend?
How do you define a trigger?
Where do you see profit?
How do you define reasonable loss?

Each and every one of these can be defined through the simplicity of Mql4 and finding ways of defining this can improve your overall understanding of the market.

IF you work in a limited frame of reference, if you have only applied a set of logic to a single pair, time frame or direction...if you have only tried to apply base logic without the understanding of the logic, you are not doomed to fail, but you limit your chances of success.

You don't need to know Java, C, C++, R, machine language...but it can't hurt.
You don't need to study the market in a live environment for thousands of hours, but it can help.
If you want to understand methodology and are limited in the time you can lend to market, the use of an automated system can help you immensely.
If you have no understanding of Mql4 but have the ability to crack open your metaeditor, you should. By gaining a simple understanding of where and when trades get taken, by understanding where something has failed, can lead you to the improvements. By applying a set of rules universally to multiple instruments and timeframes will allow you to cherry pick and improve on your probabilities for successful trading.

People who say it cannot be done should not interrupt those who are doing it.
~ George Bernard Shaw
Money Can't Buy Happiness. Poverty Can't Buy SHIT! You Choose!
  • #62
  • May 16, 2015 1:42pm May 16, 2015 1:42pm
Appreciate the input. I started with manual breakout trading and enjoyed it. The problem is I could not be awake or in front of my platform 24/5. So I started looking for ways to automate it. That is where I an now. Looking for some way to automate horizontal breakout trading.
  • #63
  • May 16, 2015 2:06pm May 16, 2015 2:06pm
Quoting Magix
{quote} This is pure bullshit. Not a single shred of truth to much of any of it, really.
Yum, another 10k+ poster. I never said or implied you couldn't use MQL4 for your research. I said you could use any tool to do data research. The keyword being data research. I also never said anything about a live environment. I don't know about you, but I wouldn't consider staring at a live market environment for thousands of hours to be a productive way to research. I'll condense what I said: You need to put real thought and research into your work to get anything but mediocre results or results that outright trick you.

Quoting Magix
If you want to understand methodology and are limited in the time you can lend to market, the use of an automated system can help you immensely.
Obviously. You just can't replace a good trader's 20 billion neuron cortex (assuming the cerebellum only does motor related tasks) with an algorithm composed of a few if statements that compare the price/some simple derivative with a few constants. I'm not a fan of trading manually myself, I could never do except on a long term basis. Selling the ruble when all of that was going on, buying oil at $54 after it had transitioned from a falling dagger, and so on. Intraday trading I would never do manually.

Quoting Magix
How do you define a trend?
How do you define a trigger?
Where do you see profit?
How do you define reasonable loss?

Each and every one of these can be defined through the simplicity of Mql4 and finding ways of defining this can improve your overall understanding of the market
Absolutely, and we all know that testing a few 50 line algorithms are the best way to do that...

Quoting Magix
If you have no understanding of Mql4 but have the ability to crack open your metaeditor, you should. By gaining a simple understanding of where and when trades get taken, by understanding where something has failed, can lead you to the improvements. By applying a set of rules universally to multiple instruments and timeframes will allow you to cherry pick and improve on your probabilities for successful trading.
Now you're just repeating what I was saying. I said that you need to study the trades taken by the algorithm. Each and every single one, preferably. Looking at not just the price chart, but at all the data the algorithm had available that made it take the respective trade.

Quoting Magix
People who say it cannot be done should not interrupt those who are doing it.
~ George Bernard Shaw
None of you are.
  • #64
  • May 16, 2015 2:19pm May 16, 2015 2:19pm
Quoting CuriosityFX
{quote} Yum, another 10k+ poster. None of you are.
Don't pretend to know me, what I do or how I do it.

You don't.

It's easy to see that you like to believe that everything fits neatly into one little package...it doesn't.

Your limited frame of reference is what distorts your optics, much in the same way as anybody new to market can rationalize through selective viewing.

So, you've tapped out.

Sorry bout your luck.

Move along.
Money Can't Buy Happiness. Poverty Can't Buy SHIT! You Choose!
  • #65
  • May 16, 2015 2:31pm May 16, 2015 2:31pm
Quoting CuriosityFX
You can not randomize your way to a profitable 'EA'. You can not stumble upon a profitable set of parameters and indicators for an 'EA'. You can not trade with "Expert Advisors" that are somehow experts without you yourself being an expert. The only profitable expert advisors you will find were 'profitable' in backtests due to chance, not due to an understanding of the market. Once you become an expert, you don't use "Expert Advisors", you use algorithms or models. The algorithms don't come from stumbling around, they come from years of research...

I agree with much of what's said here. But I want to add a few points based on my experience and understanding of the market...

I have done some research in FX using machine learning etc, by adopting a portfolio approach to the market. It's profitable and often beats the benchmarks, about 10% p.a. type of profit without leverage. heck, even simple carry trade is profitable...

But the thing is, these models make use of very limited data and really treats the data in mostly linear fashion. In trading, the decisions we make are highly non-linear. For example, setting variable TP and SL settings depending on market conditions; using jumping stops that depend on price breaking certain zones; trading only at certain times; responding to significant news events depending on how the market reacted to it in real time; not trading because price has not met some complicated condition... These are highly non-linear decisions and many of them rely on variables/facets of the market that are themselves non-linear combination of simple price, time, news information... So I wouldn't be surprised that many traders can make money and big money. The edge is definitely there... It's just difficult to exploit it consistently.

So I'm confident a good strategy can be coded and be profitable, but only if the trader was already profitable with it manually. Andrew Ng said that, a good question to ask before attempting machine learning is, if we gave these information and data to an expert in the field, will he or she be able to make good predictions? If the answer is no, then machine learning is unlikely to work... The problem is in the information we use; semi-supervised learning could work. The first step is to get an expert to pre-process the information fed to the algo.
Build good relationships with others.
  • #66
  • May 16, 2015 9:22pm May 16, 2015 9:22pm
Thanks... Garry

Quoting ibthescottyb
New thread started for EA testing only. http://www.forexfactory.com/showthre...11#post8266611
  • #67
  • May 16, 2015 9:34pm May 16, 2015 9:34pm
Thanks... Garry

Quoting ibthescottyb
New thread started for EA testing only. http://www.forexfactory.com/showthre...11#post8266611
  • #68
  • Jun 5, 2015 10:04am Jun 5, 2015 10:04am
Quoting tablet8
{quote} Hey Guys here is this robot: never trust this robots I can give u many similar robots making 1 million dollars from 30 dollar in a monnth in backtest and sometimes in demo forward test but in live account they all fail the reason is spread slippage which is fixed in demo and backtest but in live accounts it is killing every scalping robot another similar robot to this is angry beaver u can find it in mql5.com it makes 1000000 dollar in a month but in live account it help you loose faster {file} if u don't know about robot and want to buy...
Hi tablet8,

We've been forward testing FX Rogue Wave (2.6 update 1 version) on a demo account for 10 months (of course, demo is not same as live testing, but the difference would mostly affect EAs which scalp where slippage and spread differ from live to demo); so far the results are: 43% gain, 17% drawdown. (you can see the test here: http://www.forexverified.com/reports...uewave001.html) This version has survived for 10 months and's been constantly going up, something which cannot be said for many other EAs.
  • #69
  • Jun 5, 2015 10:30am Jun 5, 2015 10:30am
Quoting forexverif
{quote} Hi tablet8, We've been forward testing FX Rogue Wave (2.6 update 1 version) on a demo account for 10 months (of course, demo is not same as live testing, but the difference would mostly affect EAs which scalp where slippage and spread differ from live to demo); so far the results are: 43% gain, 17% drawdown. (you can see the test here: http://www.forexverified.com/reports...uewave001.html) This version has survived for 10 months and's been constantly going up, something which cannot be said for many other EAs.

Hi ForexVerified - I sent an email to you awhile back as I am no longer offering my EA's for sale. That particular one does very well, but it does use a Martingale approach and it can run for long periods of profitability, but it can have successive losses that can make the risk-reward ratio, not so profitable. You may remove it from your site.

Thank you!
  • #70
  • Last Post: Jun 5, 2015 11:34am Jun 5, 2015 11:34am
Quoting ibthescottyb
{quote} Hi ForexVerified - I sent an email to you awhile back as I am no longer offering my EA's for sale. That particular one does very well, but it does use a Martingale approach and it can run for long periods of profitability, but it can have successive losses that can make the risk-reward ratio, not so profitable. You may remove it from your site. Thank you!
Hi ibthescottyb, it seems we failed to get your email; we will retire the test.
0 traders viewing now