Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Do you believe in social trading performance data? 8 replies

After years in trading, do you still believe in get-rich-fast BS? 98 replies

Trading inside the spread - do you believe it? 1 reply

Do you believe in nest trading words... 0 replies

Some NFP forecasts talk about 200-360k ?? you believe ? 11 replies

Attachments: Do you believe in EA trading?
Exit Attachments

Do you believe in EA trading?

Hi,

As a relative newbie to trading i wondered if someone could
explain why in quite a lot of back tests i have done, the profit
graph keeps going up and up for years but then right at the
end plummets, sometimes to zero?
This is a response to both 1kc001 and jordanknight.

EAs WILL break down in time. Sure you might get very very lucky and have an EA that continuously is profitable for 10 or more years. But the problem is, all EAS are based off of the market conditions they were created in, and markets always tend to change at one point or another. And as soon as the current market conditions deviates too far from the market conditions the EA was created for, you will see your losses piling on.

I find it a bit shocking to see so many new traders holding EAs (or any other completely mechanical system for that matter) in such high regard considering that there has never been a single fully mechanical system that has consistently withstood the test of time.

Some of you might argue that automated and mechanical systems work because pros sometimes use them. That's certainly not false, but I have never heard of any pro trader depending on any single system to last them for their entire career. Professional system developers and traders consistently create new systems and refine older systems as market conditions change. Pros have the full and realistic expectation that whatever system they might be using at the time just might fail on them in even the next weeks and they have the foresight to switch systems before it's too late.
Quoting 1kc001
Hi,

As a relative newbie to trading i wondered if someone could
explain why in quite a lot of back tests i have done, the profit
graph keeps going up and up for years but then right at the
end plummets, sometimes to zero?
One reason could be because the tester closes any open trade at the end of the period being tested, if any negative trades going on or danglers, everything will be closed as it is.
Quoting doblece
One reason could be because the tester closes any open trade at the end of the period being tested, if any negative trades going on or danglers, everything will be closed as it is.

Thanks. What does it say about the profitability of an ea? If the ea was trading profitably in back test up until the end close out of positions, does that mean it could still be profitable?
Quoting Inzider
no.

market are changing beats, ea can help for manual trading, but any of those will blow an account on long period.

have fun!
If a method works, then it works. Period. No matter what the market conditions.

It's all about simple profit vs loss on the long run. If you have more profits than losses in the long run, the ea is profitable on the long run. Simple as that.

EAs are all about quantative, not qualitive. You will be able to lose a few trades if the rest make more profit than the losers and you are on profit. Then repeat this million times and you have much more profit.

It's not really that hard.
Quoting mikkom
If a method works, then it works. Period. No matter what the market conditions.

It's all about simple profit vs loss on the long run. If you have more profits than losses in the long run, the ea is profitable on the long run. Simple as that.

EAs are all about quantative, not qualitive. You will be able to lose a few trades if the rest make more profit than the losers and you are on profit. Then repeat this million times and you have much more profit.

It's not really that hard.
Ok, i do agree with you that certain robust automated systems might be profitable in the long run - say 10 to 20 years. But what that might mean is a couple of years of extraordinary profits, a couple years of massive drawdowns, and the rest of the years being more or less break even. This pretty much happens to any mechanical trading system, no matter how robust, because market conditions simply change all the time.

If you can deal with that extreme volatility in the short term knowing that the expected value is high in the long run, then by all means use the system. But I really doubt that unless you the most extreme form of "risk-loving" individual, you will be able to stomach a prolonged period of losses.

A good system trader understands the concept set above and will substitute their current system with new systems during unfavorable market conditions and will return to their old systems once the market returns to familiar conditions.
Quoting Lamdun
Ok, i do agree with you that certain robust automated systems might be profitable in the long run - say 10 to 20 years. But what that might mean is a couple of years of extraordinary profits, a couple years of massive drawdowns, and the rest of the years being more or less break even. This pretty much happens to any mechanical trading system, no matter how robust, because market conditions simply change all the time.
Massive drawdowns are not something you must have if you use decent risk levels but you are correct that there will be periods of even years that will have a quite flat/slightly descrending profit curve. It all goes down to how much you want to risk each trade (I predfer about 2-4% and my methods are longer term meaning min. daily bars).

So you have one of those methods where you will have a slow profit curve up with minor drawdown or flat periods (might be for months or even years as you said) but eventually the profit will always be up in let's say 5 years. What happens if you find many similar methods that are uncorelated (flat periods/minor dd's don't cluster to the same periods) and combine them to one EA?

IMHO It all goes back to quantative vs qualitive and I agree on qualitive part, I don't think that you can trade qualitive with algorithms. The profit mut come from many trades with small risk and high profit possibility.

I'm really tempted to post some very long term (20+ years) profit curves from my private EAs here but I think it's against the rules (?).
Quoting Lamdun
This is a response to both 1kc001 and jordanknight.

EAs WILL break down in time. Sure you might get very very lucky and have an EA that continuously is profitable for 10 or more years. But the problem is, all EAS are based off of the market conditions they were created in, and markets always tend to change at one point or another. And as soon as the current market conditions deviates too far from the market conditions the EA was created for, you will see your losses piling on.

I find it a bit shocking to see so many new traders holding EAs (or...
Perhaps, we should start something to keep track of all performing EAs to see whether they work in the long term. So far none of the arguments put forward here (whether they work or not) is backed by facts/statistical significance, just opinions and perception. No one could say for sure it's right or wrong.

EAs must not be created in one particular time period, as I mentioned earlier, it should work way from 1999/2000. If it does, it stands a good chance to work in the future. 2 yrs of good performance including the volatile 2008 not convincing?

Certainly, EAs could be programmed to sense changes in the market activites, but it'll require excellent programming skills and trading experience. This can possible make an EA flexible enough. So far most of the EAs fit well into the saying 'garbage in garbage out', probably that's why most of us are frustrated/loss hope on EAs
Adaptive autotrading is the future...
Quoting mikkom
I'm really tempted to post some very long term (20+ years) profit curves from my private EAs here but I think it's against the rules (?).
Where did you get 20+ years of forex data to test on? MT4/MT3/MT2/MT1 has been around for that long?
Adaptive autotrading is the future...
Quoting jordanknight
Where did you get 20+ years of forex data to test on? MT4/MT3/MT2/MT1 has been around for that long?
My methods are longer term so I don't need shorter term data - I test on daily bars.

I use 1 min data for validation (alpari has 1 min data since 2004+ or something) but just to validate that 1 min data gives me exactly or very close the same results as the daily bar data on the same period.

edit: link http://www.alpari.co.uk/en/dc/databank.html
well it has finally occured to me after all this time where people are coming from

most folks here want to become traders. and what does that mean? grab a few tools and hammer out profits.

well, i just realized that i have been a space cadet my entire FF life. instead of empowering myself, i have given the market the power. i have let her dictate my actions. i have created my systems based on her. i have not tried to fix something with a hammer - i have tried to tune my strategies to her.

my latest creation is an EA backtest 9 years for 30,000 pips on EURUSD. this does not mean they are raw, single entry. there is scaling in and out. but yes, i think that an EA will make profit if it continues keeping its historical characteristics. if your EA matches "the personality" of the pair, you should do okay.

but who is to say a pair will stop behaving like it always has? that's where EAs always fail.
What? Tdion being positive about harnessing the market? Wow things have changed since ive been gone.

Haha good to see you are still stirring the pot my friend!



Quoting tdion
well it has finally occured to me after all this time where people are coming from

most folks here want to become traders. and what does that mean? grab a few tools and hammer out profits.

well, i just realized that i have been a space cadet my entire FF life. instead of empowering myself, i have given the market the power. i have let her dictate my actions. i have created my systems based on her. i have not tried to fix something with a hammer - i have tried to tune my strategies to her.

my latest creation is an EA backtest 9 years for 30,000...
Lemming Forex! The Way to Trade....
Quoting Lamdun
But what that might mean is a couple of years of extraordinary profits, a couple years of massive drawdowns, and the rest of the years being more or less break even. This pretty much happens to any mechanical trading system, no matter how robust, because market conditions simply change all the time.
That's true, the idea of EA is to design it in such way so that there are more than 1 system in it, working in various conditions independently. It may be designed so that each system is triggered by certain criteria or all of the system working simultaneously to reduce the overall drawdown. The problem with most EAs is that they only use 1 system that trades only that particular type of market (eg trending) and fails when market does not behave that way.

This type of EAs are not easy to code but it is certainly possible.
Adaptive autotrading is the future...
  • Post #54
  • Quote
  • Last Post: Feb 9, 2009 2:34pm Feb 9, 2009 2:34pm
i good post and strategy i implore myself.

but, all "systems" can draw down at the same time.

Quoting jordanknight
That's true, the idea of EA is to design it in such way so that there are more than 1 system in it, working in various conditions independently. It may be designed so that each system is triggered by certain criteria or all of the system working simultaneously to reduce the overall drawdown. The problem with most EAs is that they only use 1 system that trades only that particular type of market (eg trending) and fails when market does not behave that way.

This type of EAs are not easy to code but it is certainly possible.
0 traders viewing now