the true unprofitable face of the adviser - page 2

 
delyus:

The only Expert Advisor that showed profit in the Strategy Tester, with quotes downloaded from the Center's Hitory, immediately plunged the deposit into the floor.


I had the opposite situation. A fairly good EA showing acceptable results on broker's quotes has yielded nearly smooth results on the histograms: 3 years, 1700 trades, the curve, or rather a nearly straight line with the angle of 40 degrees. And that doesn't tell you anything, just like your negative result. The chistoricenter quotes, although not fundamentally, but they are different from your broker's quotes and this is often enough for serious differences in the results. It is better to test and optimize the Expert Advisor on the broker's quotes which you are going to use it with. The quality of modeling, imho, does not play a big role in assessing the results of the expert, because even at 90% it is only one of the possible options of price behavior, not worse or better than others. But if this very quality noticeably affects the result of testing, then I think it is a reason to think about the stability of the trading system and its safe use in real trading.
 
DrawDown:
We should test and optimize an Expert Advisor, that does not depend on intra-momentary price fluctuations, i.e. stop at least 50 points and take at least 150 points. Then there will be the same result whether on brokerage prices or on prices of 5-10 brokerage companies, or on real prices of the same brokerage companies. And if it shows profit within 7 years of back-testing, then there is a chance (but it is low) that it will be almost the same on forward testing, even if the forward test is carried out on a real account. Otherwise, a dumb loss-making EA can make 250-300% during a year and then lose everything in a month. Among 2-3 thousand testers on the same Viac, there are those who made so much money during the first year of the test, but could not keep the deposit for the next or third month after that. The same will be at the second championship... The first one was the same, and in the contest of brokerage companies only the crazy trader that opened one side of one pair for the entire deposit and hit the bull's-eye. And there are a lot of them among several thousands, that's why they have more chances. Of course, I'm exaggerating, but I hope the point is clear.

About the intra-bar (not necessarily intra-minute) movements - I agree. It should be based on OHLC. But concerning the coincidence of the results when testing on quotes of different brokers - is questionable. For example, for the presence/absence of a fractal signal a difference of only 1 point is sufficient.
 
Analitik:
DrawDown:
It is necessary to test and optimize an Expert Advisor which does not depend on intra-momentary price fluctuations, i.e. stop at least 50 points and take at least 150 points. Then either on brokerage quotes or on quotes of 5-10 brokerage companies, or on real prices of the same brokerage companies - the result will be the same. And if back-testing shows profit within 7 years there is a chance (but it is very low) that it will be almost the same in forward testing even if the forward test is carried out on the real account. Otherwise, a dumb loss-making EA can make 250-300% during a year and then lose everything in a month. Among 2-3 thousand testers on the same Viac, there are those who have made so much money during the first year of the test and then failed to keep the deposit for the next or third month after that. The same will be at the second championship... The first one was the same, and in the contest of brokerage companies only the crazy trader that opened one side of one pair for the entire deposit and hit the bull's-eye. And there are a lot of them among several thousands, that's why they have more chances. Of course, I am exaggerating, but I hope the point is clear.

About the intra-bar (not necessarily intra-minute) movements - I agree. It should be based on OHLC. But concerning the coincidence of the results when testing on quotes of different brokers - is questionable. It depends on the signals that underlie the strategy. For example, for the presence/absence of a fractal signal a difference of only 1 point is sufficient.

If 1 point makes the weather, then we cannot speak about the independence of the Expert Advisor from the intra-bar movements. And I believe the presence or absence of the fractal signal should not be determined point by point, but the range of 3-7 points and shift it relative to the actual value of the fractal, depending on the trade direction. Then test results of different brokerage companies will coincide. And if there is no desire to lose these 3-7 points, it means that the Expert Advisor depends on intra-bar movements.
 
DrawDown:

If 1 point makes the weather, then we cannot talk about the independence of the Expert Advisor from intra-bar movements. And I believe that the presence/absence of a fractal signal should not be determined point by point, but set a range of 3-7 points and shift it relative to the actual value of the fractal, depending on the direction of the deal. Then the test results of different brokerage companies will be the same. And if there is no desire to lose these 3-7 points, it means that the Expert Advisor depends on the intra-bar movement.

But this is a delusion. The price will pass 7 points by one brokerage company from the fractal level and the signal will work. In another brokerage company the price will pass through 8 points and the signal will not work. As the result different brokerage companies will have different testing results. I.e. the notorious effect of 1 point will not disappear.
 

Many people know how much I struggle with EAs. Again, I have tested over 2000 EAs on all ticks. I have found one profitable one - TSD group (Alpari, 7 years, 5 thousand, 3 lots, 70 000). On his motivations I have made a new EA with Valmars (7 years, 5 thousands, 3 lots, 250 000). The same EA has shown on Vhc (7 years, 5 thousands, 3 lots, 650 000). As soon as I downloaded quotes from NС the securities were lost, however roughly I set them. Only TSD with 50 000 depo somehow managed to survive 1 lot in 7 years, showing a measly 70 000 - 20 000 profits with huge drawdowns.

I just want to know, has anyone ever tested a profitable Expert Advisor from 2000 till now on all ticks or on opening prices according to quotes from NS? Is it possible in principle? If there are any, please post the statement for that period.

 
bstone:
DrawDown:

If 1 point makes the weather, then you can no longer talk about the independence of the EA from intra-bar movements. And I believe that the presence/absence of a fractal signal should not be determined point by point, but set a range of 3-7 points and shift it relative to the actual value of the fractal depending on the direction of the trade. Then the test results of different brokerage companies will be the same. And if there is no desire to lose these 3-7 points, it means that the Expert Advisor depends on intra-bar moves again.

But this is a delusion. The price will pass 7 points by one brokerage company from the fractal level and the signal will work. In another brokerage company the price will pass through 8 points and the signal will not work. As the result different brokerage companies will have different testing results. I.e. the notorious effect of the 1st pip will not disappear.

This is an example, though rough, but the essence is clear. If difference of one pip at different brokerage companies influences the final result, this is not the Expert Advisor that we may use in real trading.
 
DrawDown:

This is just a rough example, but I think the point is clear. If the difference of one point in different brokerage companies has an impact on the final result, then this is not an EA that can be used in real trading.

I thought so too. But not so long ago I built an EA implementing the strategy that one of my friend has been successfully trading for three years on a real account through IGM. I ran this EA on a period of real trading with specified settings but using Riston Capital's quotes - the deposit has been lost for 6 months. We analyzed each deal. The Expert Advisor worked absolutely correctly. The reason was those few (and sometimes not few) points of quotes difference.
 
delyus:

Many people know how much I struggle with EAs. Again, I have tested over 2000 EAs on all ticks. I have found one profitable one - TSD group (Alpari, 7 years, 5 thousands, 3 lots, 70 000). On his motivations I have made a new EA with Valmars (7 years, 5 thousands, 3 lots, 250 000). The same EA has shown on Vhc (7 years, 5 thousands, 3 lots, 650 000). As soon as I downloaded quotes from CS the deposit was immediately lost, however roughly I adjusted it. Only TSD with 50 000 depo somehow managed to survive 1 lot in 7 years, showing a measly 70 000 - 20 000 profits with huge drawdowns.


I just want to know, has anyone ever tested a profitable Expert Advisor from 2000 till now on all ticks or on opening prices according to quotes from NS? Is it possible in principle? If there are any, please post the statement for that period.



Will this do?
 
Analitik:
DrawDown:

This is just a rough example, but I think the point is clear. If the difference of one point in different DCs has an impact on the final result, then this is not an EA that can be used in real time.

I thought so too. But not so long ago I built an EA implementing the strategy which one of my friends has been successfully trading for three years on real account through IGM. I ran this Expert Advisor on a period of real trading with its settings but using Riston Capital's quotes, the deposit has been lost in 6 months. Didn't feel lazy - took apart every trade. The reason was those few (sometimes not few) points of quotes difference.

Thus, nothing prevents IGM to bring its quotes to such a condition that the Expert Advisor of your friend (and not only his) will trade worse than Riston Capital's ones (though the difference between them is barely noticeable). If the Expert Advisor is sensitive to quotes, the slightest changes in the market will affect its balance. Your friend will notice it only when the Expert Advisor starts to lose.
 

conys,

I'm shocked, is it really possible? It can't be! He's making profit on the minutes! He trades somewhere in the real world? He seems to be a very rough setup and a Pipsman at the same time.