My EA's test results - what's the catch and illusion? - page 2

New comment
[Deleted]  
Figuring out how a tester works is 100500 times easier than figuring out how to trade in the market (I don't know how to trade), but the tester is the only way to test any brilliant strategy without spending a lot of money and most importantly time
 
concord99:

The thing is, I'm not a pipsitter in terms of loss-profit ratio.

And in terms of setting a stop?

 
concord99:

Good afternoon.


What kind of orders is the strategy based on? (market, limit, stop orders?)
 
PapaYozh:

And in terms of setting the stop?

I generally understand piping as wanting to take 1-2 pips. My take is clearly bigger. A short stop loss in my understanding is not a sign of pipsing. Although perhaps there are other definitions of piping nowadays.
 
DYN:

Strategy on which orders? (market, limit, stop-loss orders?)

Afternoon. I use pending stop orders.
 
concord99:

Good afternoon.

I guess a lot of people are already going on holiday etc. But I hope some experienced TRADERS will be able to give critical comments on the testing results of my Expert Advisor for 2007-2013. I'm sceptical about it myself, I expect experienced people to advise - what's the catch of the next variant. I think that's the end of my attempts to invent the "grail" :)

The lot size did not change within the testing period. Stepwise increase of lot size leads to "flight into space" - tens of millions of dollars :)))

Thanks in advance to all who will write on the merits.

If you are testing the strategy on a large timeframe, for example on H1, and you want maximum accuracy, and you need it because the stop size is extremely small, then you should use the "Fast method on formed bars" mode. Only for Expert Advisors with explicit capital control". You transfer the logic to M1, and work out stops and takeprofits only at existing M1 prices - i.e. at OHLC, i.e. you have to replace stoploss and takeprofit orders with market orders. If you need a stop within the range of M1 bars, then it is more difficult to do so correctly, as MT4 does not correctly reflect the triggering of stops on gaps. Let us assume that the testing mode is "fast method". At the opening of a new bar you set the stop loss equal to 20 points of the price. Within a minute the price trades without reaching this stop loss. Then a gap in time follows, say a weekend. And the opening price of the next bar is 100 points below the previous bar's close. In this case your stop will be triggered, at the previous -20 pips, though it cannot be triggered there, as there are no prices there. This should be taken into account, especially for instruments with low liquidity and frequent gaps.
 
concord99:
I generally understand piping as wanting to take 1-2 pips. My take is clearly bigger. A short stop loss is not a sign of pipsing in my mind. Although perhaps there are other definitions of pips nowadays.
An ultra-short stop is almost impossible in a robust strategy. It is incredibly difficult to put just a short stop, while an ultra-short stop is a miracle of miracles. There is a simple test to find out if you are using a tester or market pattern, to do this simply remove the stop and takeprofit and use time out instead. If the results become dramatically different for the worse - congratulations, you've found a way to outsmart the tester. But it will be of little use, because the market will not be circumvented that way).
[Deleted]  
C-4:
If you're testing the strategy on a large timeframe, such as H1, and you need maximum accuracy, and you need it because the stop size is very small, then you need to use the "Fast method on formed bars" mode. Only for Expert Advisors with explicit capital control". You transfer the logic to M1, and work out stops and takeprofits only at existing M1 prices - i.e. at OHLC, i.e. you have to replace stoploss and takeprofit orders with market orders. If you need a stop within the range of M1 bars, then it is more difficult to do so correctly, as MT4 does not correctly reflect the triggering of stops on gaps. Let us assume that the testing mode is "fast method". At the opening of a new bar you set the stop loss equal to 20 points of the price. Within a minute the price trades without reaching this stop loss. Then a gap in time follows, say a weekend. And the opening price of the next bar is 100 points below the previous bar's close. In this case your stop will be triggered, at the previous -20 pips, though it cannot be triggered there, as there are no prices there. This should be taken into account, especially for instruments with low liquidity and frequent gaps.


This gap may work out differently in real life (and in the tester, it's a goddamn plus).

By the way, if you use any indicator, you may also leave some comments here on how to prepare it correctly for the Strategy Tester.



 
C-4:
If you are testing the strategy on a large timeframe, e.g. H1, and you need maximum accuracy, which you do because the stop size is extremely small, then you should use the "Fast method on formed bars" mode. Only for Expert Advisors with explicit capital control". You transfer the logic to M1, and work out stops and takeprofits only at existing M1 prices - i.e. at OHLC, i.e. you have to replace stoploss and takeprofit orders with market orders. If you need a stop within the range of M1 bars, then it is more difficult to do so correctly, as MT4 does not correctly reflect the triggering of stops on gaps. Let us assume that the testing mode is "fast method". At the opening of a new bar you set the stop loss equal to 20 points of the price. Within a minute the price trades without reaching this stop loss. Then a gap in time follows, say a weekend. And the opening price of the next bar is 100 points below the previous bar's close. In this case your stop will be triggered, at the previous -20 pips, though it cannot be triggered there, as there are no prices there. This should be considered especially for instruments with low liquidity and frequent gaps.


Thank you. I will try to figure it out. But unfortunately I am less and less optimistic ).... I am a skeptic myself, that is why I try to quickly debunk my own grail program))) ....
 
YOUNGA:

this gap can work in real life (but in a tester it's a goddamn plus)

And as for the stops, the gap is always in the plus, which may lead to another pseudo-grade
12345678
New comment