Behemoth Eur/Usd 1 Lot per trade (由 stevetrade)
該用戶已將其策略轉為私有。
Behemoth Eur/Usd 1 Lot per trade 討論
Jun 24, 2010 at 13:30
1,364流覽
9 Replies
會員從Oct 28, 2009開始
1430帖子
Jun 24, 2010 at 13:39
(已編輯Jun 24, 2010 at 13:39)
This is version 0.2 of Behemoth.
It's still in Alpha development stage.
Again, modelling quality is n/a as I am using fixed range bars, which always return n/a.
I hope to have this forward testing by the end of the week.
It's still in Alpha development stage.
Again, modelling quality is n/a as I am using fixed range bars, which always return n/a.
I hope to have this forward testing by the end of the week.
11:15, restate my assumptions: 1. Mathematics is the language of nature. 2. Everything around us can be represented and understood through numbers. 3. If you graph these numbers, patterns emerge. Therefore: There are patterns everywhere in nature.
forex_trader_7
會員從Aug 01, 2009開始
941帖子
Jun 24, 2010 at 13:59
Bud,
Those tests are useless. Even with tick by tick modeling they'll be out many, many percentage points. At best they can be used to see if the code vaguely does what it's supposed to do.
It's a complete waste of time.
Those tests are useless. Even with tick by tick modeling they'll be out many, many percentage points. At best they can be used to see if the code vaguely does what it's supposed to do.
It's a complete waste of time.
會員從Oct 28, 2009開始
1430帖子
Jun 24, 2010 at 14:01
As I said in my first post on this EA. Only forward testing will tell if this EA really works.
It's the limitation of using fixed range bars. They always return n/a as MT4 classes all the bars as mismatched chart errors due to the changing timeframe.
It's the limitation of using fixed range bars. They always return n/a as MT4 classes all the bars as mismatched chart errors due to the changing timeframe.
11:15, restate my assumptions: 1. Mathematics is the language of nature. 2. Everything around us can be represented and understood through numbers. 3. If you graph these numbers, patterns emerge. Therefore: There are patterns everywhere in nature.
forex_trader_7
會員從Aug 01, 2009開始
941帖子
Jun 24, 2010 at 15:22
Limitation? They don't even vaguely reflect reality.
I must say I completely fail to see the value of the strategy section. To me it's based om something that's fundamentally flawed. All it does is creates confusion and maybe even give back testing some form of credibility.
Different machines will give you different results, different processors will give you different results, if you're online or not will give you different results, a different broker will give you different results... I have yet to see a days trading match a back test for the same day.
All a load of crock. Only way to test is by testing live as you said, but event that will differ from a real live account. So this lot is a complete waste of time in my humble opinion.
I must say I completely fail to see the value of the strategy section. To me it's based om something that's fundamentally flawed. All it does is creates confusion and maybe even give back testing some form of credibility.
Different machines will give you different results, different processors will give you different results, if you're online or not will give you different results, a different broker will give you different results... I have yet to see a days trading match a back test for the same day.
All a load of crock. Only way to test is by testing live as you said, but event that will differ from a real live account. So this lot is a complete waste of time in my humble opinion.
會員從Oct 28, 2009開始
1430帖子
Jun 24, 2010 at 15:30
(已編輯Jun 24, 2010 at 15:33)
You are of course at liberty to completely ignore this section then.
For those of us that do think that there are benefits to backtesting and are aware of it's limitations it's been a godsend.
For those of us that do think that there are benefits to backtesting and are aware of it's limitations it's been a godsend.
11:15, restate my assumptions: 1. Mathematics is the language of nature. 2. Everything around us can be represented and understood through numbers. 3. If you graph these numbers, patterns emerge. Therefore: There are patterns everywhere in nature.
forex_trader_7
會員從Aug 01, 2009開始
941帖子
Jun 24, 2010 at 16:11
(已編輯Jun 24, 2010 at 16:28)
Just surprised to find you here.
Personally I think it does harm to myfxbook, or anyone else that partakes in the section. Only a complete newby or ignoramus would go to the hassle of analyzing a back test.
The data it's based on is flawed. Unless you own a bank you just can't do it.
Let me tell you a story, when I was getting into automated fx my partner was programming for an engineering company, more to get residency than anything else.
We got into developing and the boss of the engineering firm heard of this he decided he was going to have a go, we even had a meeting to see if we could do it for him, and then he went on to hire some hotshot young programmer come economist from overseas to come do the job.
This guy calculated a system, then calculated the probability of his system being a working one. He calculated he had an 80% chance that his system will work. They put it in the market - no test at all - and got nuked to the tune of 6 zeros' in a matter of days.
And back testing is like that to me, it's a calculation on a calculation on a calculation.
Personally I think it does harm to myfxbook, or anyone else that partakes in the section. Only a complete newby or ignoramus would go to the hassle of analyzing a back test.
The data it's based on is flawed. Unless you own a bank you just can't do it.
Let me tell you a story, when I was getting into automated fx my partner was programming for an engineering company, more to get residency than anything else.
We got into developing and the boss of the engineering firm heard of this he decided he was going to have a go, we even had a meeting to see if we could do it for him, and then he went on to hire some hotshot young programmer come economist from overseas to come do the job.
This guy calculated a system, then calculated the probability of his system being a working one. He calculated he had an 80% chance that his system will work. They put it in the market - no test at all - and got nuked to the tune of 6 zeros' in a matter of days.
And back testing is like that to me, it's a calculation on a calculation on a calculation.
會員從Oct 28, 2009開始
1430帖子
Jun 24, 2010 at 16:24
(已編輯Jun 24, 2010 at 16:25)
Yes, I agree. It's very easy to fool yourself into thinking that your system works because of a backtest and putting it live on serious money before it looks good on small money is just silly.
However, I think if you keep the concepts simple and work from sounds trading principles then there is less to go wrong.
Most of the issues with backtesting come from known problems, such as the strategy tester using a fixed spread on the entire test or price spikes that aren't included in the tick data on the strategy tester, but these things can be catered for both in your strategy and in your approach to coding.
As a starting point on which to test the logic of a trading idea. It can't actually be beaten, because there is no other way to do it other than forward trading or god forbid manually backtesting, and that obviously takes time.
So, what would you rather do? Start with a strategy based on solid trading ideas, that has a good backtest and forward test it or start blind with a strategy that looked as though it worked on a few days data, or as much as you can stomach backtesting manually?
However, I think if you keep the concepts simple and work from sounds trading principles then there is less to go wrong.
Most of the issues with backtesting come from known problems, such as the strategy tester using a fixed spread on the entire test or price spikes that aren't included in the tick data on the strategy tester, but these things can be catered for both in your strategy and in your approach to coding.
As a starting point on which to test the logic of a trading idea. It can't actually be beaten, because there is no other way to do it other than forward trading or god forbid manually backtesting, and that obviously takes time.
So, what would you rather do? Start with a strategy based on solid trading ideas, that has a good backtest and forward test it or start blind with a strategy that looked as though it worked on a few days data, or as much as you can stomach backtesting manually?
11:15, restate my assumptions: 1. Mathematics is the language of nature. 2. Everything around us can be represented and understood through numbers. 3. If you graph these numbers, patterns emerge. Therefore: There are patterns everywhere in nature.
forex_trader_7
會員從Aug 01, 2009開始
941帖子
Jun 24, 2010 at 16:43
No back testing, manually or otherwise.
I have my idea, we write it, then I run the code as a back test on tick by tick to see if the EA does what it's supposed to do. If it looks roughly right I ignore the results, because there is not foundation in them and then I go live, where most of the time I'll find bugs anyway.
I have no idea how many working systems are in the dustbin because of back testing, but I'd bet it's a lot.
What I'd rather do is open different MT4 directories on the same server and run them all live demo. Easily have 10 running at a time.
I have no interest in the back testing, it's nothing more than a really wild MT4 guess...how many of your systems have you binned on a back test? Do you really know if they worked or not? Or did you go ahead and test them all anyway? I do. I do not believe for even a second the data we have can give us accurate results. Not on MT4.
I have my idea, we write it, then I run the code as a back test on tick by tick to see if the EA does what it's supposed to do. If it looks roughly right I ignore the results, because there is not foundation in them and then I go live, where most of the time I'll find bugs anyway.
I have no idea how many working systems are in the dustbin because of back testing, but I'd bet it's a lot.
What I'd rather do is open different MT4 directories on the same server and run them all live demo. Easily have 10 running at a time.
I have no interest in the back testing, it's nothing more than a really wild MT4 guess...how many of your systems have you binned on a back test? Do you really know if they worked or not? Or did you go ahead and test them all anyway? I do. I do not believe for even a second the data we have can give us accurate results. Not on MT4.
forex_trader_7
會員從Aug 01, 2009開始
941帖子
Jun 24, 2010 at 16:56
And I must tell you. I been looking at the HFT stuff recently, and market makers, and I wonder if MT4 is just a really, really clever way of front running every amateur fx trader in the world...
Someone fairly recently pointed out to me that we don't really have any idea what MT is sending out. To the point where it's debatable whether the EA's get sent out.
Someone fairly recently pointed out to me that we don't really have any idea what MT is sending out. To the point where it's debatable whether the EA's get sent out.
會員從Oct 28, 2009開始
1430帖子
Jun 24, 2010 at 19:53
(已編輯Jun 24, 2010 at 19:54)
I've binned countless systems. Some because they were flawed ideas, which were shown up by a visual backtest but others because backtesting identified that there were periods where a currency range traded for an extended period and the EA would have caused unacceptable drawdown - or similar changes to a currency pairs behavior were flagged up.
I've heard the theory that the EA's get sent out to the brokers before. I think it's pure paranoia to think that the entire Forex world is being driven by a load of retail investors using Expert Advisors. They would need some giant supercomputer somewhere that logs who is using which EA, how big their trades are and what the best permutation is to make the most money off the traders. It's not likely. I think they just rely on the fact that the majority of people are incompetent and don't do their research properly.
I frequently check my MT4 broker prices against my Dukascopy account when I think something weird has happened on MT4. Invariably I find that the price that the market maker gave me is pretty much the same as Dukascopy had at the same time. Now, I'm not saying that all brokers are like that, I'm well aware that there are some MT4 brokers out there that manipulate the feed, but if you stay within the confines of the well known market makers then things will work out pretty much as the market dictates regardless of which EA you are running.
Yes you might get disconnected or requoted from time to time but there are ways around that as well.
I've heard the theory that the EA's get sent out to the brokers before. I think it's pure paranoia to think that the entire Forex world is being driven by a load of retail investors using Expert Advisors. They would need some giant supercomputer somewhere that logs who is using which EA, how big their trades are and what the best permutation is to make the most money off the traders. It's not likely. I think they just rely on the fact that the majority of people are incompetent and don't do their research properly.
I frequently check my MT4 broker prices against my Dukascopy account when I think something weird has happened on MT4. Invariably I find that the price that the market maker gave me is pretty much the same as Dukascopy had at the same time. Now, I'm not saying that all brokers are like that, I'm well aware that there are some MT4 brokers out there that manipulate the feed, but if you stay within the confines of the well known market makers then things will work out pretty much as the market dictates regardless of which EA you are running.
Yes you might get disconnected or requoted from time to time but there are ways around that as well.
11:15, restate my assumptions: 1. Mathematics is the language of nature. 2. Everything around us can be represented and understood through numbers. 3. If you graph these numbers, patterns emerge. Therefore: There are patterns everywhere in nature.
*商業用途和垃圾郵件將不被容忍，並可能導致帳戶終止。
提示：發佈圖片/YouTube網址會自動嵌入到您的帖子中！
提示：鍵入@符號，自動完成參與此討論的用戶名。
