Looks promising. However I was only able to conduct a backtest for EURUSD "every tick mode". When trying "Control points" or "Open price" and/or GBPUSD I get a lot of errors e.g.
-> "critical error in the EA" -> "zero divide in 'Steed.mq4' (2997,52)" -> "array out of range in 'Steed.mq4' (3505,25)"
I tested with FXCM (commission type account) and "Admiral.Prime" (ECN type account @ Admiral Markets).
Looks promising. However I was only able to conduct a backtest for EURUSD "every tick mode". When trying "Control points" or "Open price" and/or GBPUSD I get a lot of errors e.g.
-> "critical error in the EA" -> "zero divide in 'Steed.mq4' (2997,52)" -> "array out of range in 'Steed.mq4' (3505,25)"
I tested with FXCM (commission type account) and "Admiral.Prime" (ECN type account @ Admiral Markets).
What's wrong here?
Hi!
Most likely the problem is the lack of quotes. The EA analyzes several timeframes, so it is necessary to download the history of quotes for all timeframes (at least for three months before the start date).
Please let me know if the problem is not resolved.
Most likely the problem is the lack of quotes. The EA analyzes several timeframes, so it is necessary to download the history of quotes for all timeframes (at least for three months before the start date).
Please let me know if the problem is not resolved.
I'm using HQ data (full price history from Dukascopy). Sometimes it works, sometimes I get said error. It seems the code is not stable. Anyway: The trading logic maybe good.
Bernhard Stremayr: I'm using HQ data (full price history from Dukascopy). Sometimes it works, sometimes I get said error. It seems the code is not stable. Anyway: The trading logic maybe good.
You're wrong. The code is stable. Do you use any third-party testing software?
Bernhard Stremayr: I'm using HQ data (full price history from Dukascopy). Sometimes it works, sometimes I get said error. It seems the code is not stable. Anyway: The trading logic maybe good.
Now I'm special downloaded a new terminal (and all history data) and tried to test the expert on GBPUSD. There were no errors.
Bernhard Stremayr: I tested both: Original MT4 resp. terminal [FXCM, commission type account model] and TDS2. In both cases I sometimes get said errors.
I also tested in TDS 2. Did you change this setting? If not, then do it. The expert needs data for the last three months.
I also tested in TDS 2. Did you change this setting? If not, then do it.
Yes, when setting this value to 1000 it works properly. I guess the same applies for live trading meaning every EA user has to ensure to have at least 1000 "H4 bars" in the chart before starting the expert on it and that's way more than your recommended "3 months of history".
I also want to mention: Basically this is a very professional system.
The advantage of using higher timeframes of course is that the
respective strategies are less spread sensitive and saver. On the other
hand there's a big disadvantage: If only a few hundred backtest trades per symbol/timeframe can be
conducted because of lacking history data (accurate intra-day data
before the year 2000 is almost impossible to find) it's hard to ensure
that a optimized system is not curve fitted (even when not being done
intentionally). So we need a lot of forward testing first...
Bernhard Stremayr: Yes, when setting this value to 1000 it works properly. I guess the same applies for live trading meaning every EA user has to ensure to have at least 1000 "H4 bars" in the chart before starting the expert on it and that's way more than your recommended "3 months of history".
No, in fact, approximately 300 H4 bars (which include 1200 H1 bars) are enough. I set an overestimated value.
Bernhard Stremayr:
I also want to mention: Basically this is a very professional system.
The advantage of using higher timeframes of course is that the
respective strategies are less spread sensitive and saver. On the other
hand there's a big disadvantage: If only a few hundred backtest trades per symbol/timeframe can be
conducted because of lacking history data (accurate intra-day data
before the year 2000 is almost impossible to find) it's hard to ensure
that a optimized system is not curve fitted (even when not being done
intentionally). So we need a lot of forward testing first...
It's not a scalper to have thousands of trades (especially on this timeframe).
But I understand you. I can only say that I did not optimize this expert. But these are only my words. You can wait and see the results.
PS: To be precise, you can find intraday data for 1999, but not earlier.
PPS: I tested the EA from 1971 (in very rough mode without using intraday data). It was profitable. If you are interested, I can show results in a private message.
No, in fact, approximately 300 H4 bars (which include 1200 H1 bars) are enough. I set an overestimated value.
It's not a scalper to have thousands of trades (especially on this timeframe).
But I understand you. I can only say that I did not optimize this expert. But these are only my words. You can wait and see the results.
PS: To be precise, you can find intraday data for 1999, but not earlier.
PPS: I tested the EA from 1971 (in very rough mode without using intraday data). It was profitable. If you are interested, I can show results in a private message.
If all supported pairs use exactly the same settings within the EA I'd
really be much impressed. However then I wonder why the user has to
specifiy the symbol that should be traded in the expert's properties
dialog box.
If all supported pairs use exactly the same settings within the EA I'd
really be much impressed. However then I wonder why the user has to
specifiy the symbol that should be traded in the expert's properties
dialog box.
No, I didn't say that. The settings differ. I meant that I have not optimized the expert, to get the curve up. In fact, I doubt that it is possible to optimize expert for more than 45 years (since 1971).
There's nothing wrong with optimization as long as you follow some rules. For example you can use in sample / out of sample "forward testing" after optimization and/or ensure to use an amount of (core) variables within the strategy that corresponds to a reasonable number of "test trades. When e.g. having 20 variables (input parameters) but only 100 "backtest trades" even the best equity curve is useless resp. can't tell you anything about your system's real profitability. However when having e.g. 6 (core) variables and 4000 (backtest) trades with a smooth equity curve chances are good to have developed a stable strategy. That's all a matter of statistics and their evaluation. In your case everything is fine apart the fact that you cant' conduct more "test trades". That's not your fault, it's simply an issue due to insufficient price data. In any case I wish you good luck and that your forward test confirms the backtests.
Bernhard Stremayr: There's nothing wrong with optimization as long as you follow some rules. For example you can use in sample / out of sample "forward testing" after optimization and/or ensure to use an amount of (core) variables within the strategy that corresponds to a reasonable number of "test trades. When e.g. having 20 variables (input parameters) but only 100 "backtest trades" even the best equity curve is useless resp. can't tell you anything about your system's real profitability. However when having e.g. 6 (core) variables and 4000 (backtest) trades with a smooth equity curve chances are good to have developed a stable strategy. That's all a matter of statistics and their evaluation. In your case everything is fine apart the fact that you cant' conduct more "test trades". That's not your fault, it's simply an issue due to insufficient price data. In any case I wish you good luck and that your forward test confirms the backtests.
Hi How many capital do i need if i run all 19 pairs in one go? thanks
Hi How many capital do i need if i run all 19 pairs in one go? thanks
Hi!
If we risk 2% per trade - $1000 and higher (or $10 and higher, if a cent account is used).
Consolidated report (Risk = 2%)
All backtests with 99% modelling quality and variable spread are attached.
Looks promising. However I was only able to conduct a backtest for EURUSD "every tick mode". When trying "Control points" or "Open price" and/or GBPUSD I get a lot of errors e.g.
-> "critical error in the EA"
-> "zero divide in 'Steed.mq4' (2997,52)"
-> "array out of range in 'Steed.mq4' (3505,25)"
I tested with FXCM (commission type account) and "Admiral.Prime" (ECN type account @ Admiral Markets).
What's wrong here?
Looks promising. However I was only able to conduct a backtest for EURUSD "every tick mode". When trying "Control points" or "Open price" and/or GBPUSD I get a lot of errors e.g.
-> "critical error in the EA"
-> "zero divide in 'Steed.mq4' (2997,52)"
-> "array out of range in 'Steed.mq4' (3505,25)"
I tested with FXCM (commission type account) and "Admiral.Prime" (ECN type account @ Admiral Markets).
What's wrong here?
Hi!
Most likely the problem is the lack of quotes. The EA analyzes several timeframes, so it is necessary to download the history of quotes for all timeframes (at least for three months before the start date).
Please let me know if the problem is not resolved.Hi!
Most likely the problem is the lack of quotes. The EA analyzes several timeframes, so it is necessary to download the history of quotes for all timeframes (at least for three months before the start date).
Please let me know if the problem is not resolved.I'm using HQ data (full price history from Dukascopy). Sometimes it works, sometimes I get said error. It seems the code is not stable. Anyway: The trading logic maybe good.
You're wrong. The code is stable. Do you use any third-party testing software?
I'm using HQ data (full price history from Dukascopy). Sometimes it works, sometimes I get said error. It seems the code is not stable. Anyway: The trading logic maybe good.
Now I'm special downloaded a new terminal (and all history data) and tried to test the expert on GBPUSD. There were no errors.
Now I'm special downloaded a new terminal (and all history data) and tried to test the expert on GBPUSD. There were no errors.
I tested both: Original MT4 resp. terminal [FXCM, commission type account model] and TDS2. In both cases I sometimes get said errors.
I also tested in TDS 2. Did you change this setting? If not, then do it. The expert needs data for the last three months.
I also tested in TDS 2. Did you change this setting? If not, then do it. The expert needs data for the last three months.
For example, if you decided to start the test from 01/01/2016, then the data should be available at least from 01/10/2015.
I also tested in TDS 2. Did you change this setting? If not, then do it.
I also want to mention: Basically this is a very professional system. The advantage of using higher timeframes of course is that the respective strategies are less spread sensitive and saver. On the other hand there's a big disadvantage: If only a few hundred backtest trades per symbol/timeframe can be conducted because of lacking history data (accurate intra-day data before the year 2000 is almost impossible to find) it's hard to ensure that a optimized system is not curve fitted (even when not being done intentionally). So we need a lot of forward testing first...
Yes, when setting this value to 1000 it works properly. I guess the same applies for live trading meaning every EA user has to ensure to have at least 1000 "H4 bars" in the chart before starting the expert on it and that's way more than your recommended "3 months of history".
No, in fact, approximately 300 H4 bars (which include 1200 H1 bars) are enough. I set an overestimated value.
I also want to mention: Basically this is a very professional system. The advantage of using higher timeframes of course is that the respective strategies are less spread sensitive and saver. On the other hand there's a big disadvantage: If only a few hundred backtest trades per symbol/timeframe can be conducted because of lacking history data (accurate intra-day data before the year 2000 is almost impossible to find) it's hard to ensure that a optimized system is not curve fitted (even when not being done intentionally). So we need a lot of forward testing first...
It's not a scalper to have thousands of trades (especially on this timeframe).
But I understand you. I can only say that I did not optimize this expert. But these are only my words. You can wait and see the results.
PS: To be precise, you can find intraday data for 1999, but not earlier.
PPS: I tested the EA from 1971 (in very rough mode without using intraday data). It was profitable. If you are interested, I can show results in a private message.
No, in fact, approximately 300 H4 bars (which include 1200 H1 bars) are enough. I set an overestimated value.
It's not a scalper to have thousands of trades (especially on this timeframe).
But I understand you. I can only say that I did not optimize this expert. But these are only my words. You can wait and see the results.
PS: To be precise, you can find intraday data for 1999, but not earlier.
PPS: I tested the EA from 1971 (in very rough mode without using intraday data). It was profitable. If you are interested, I can show results in a private message.
If all supported pairs use exactly the same settings within the EA I'd really be much impressed. However then I wonder why the user has to specifiy the symbol that should be traded in the expert's properties dialog box.
If all supported pairs use exactly the same settings within the EA I'd really be much impressed. However then I wonder why the user has to specifiy the symbol that should be traded in the expert's properties dialog box.
There's nothing wrong with optimization as long as you follow some rules. For example you can use in sample / out of sample "forward testing" after optimization and/or ensure to use an amount of (core) variables within the strategy that corresponds to a reasonable number of "test trades. When e.g. having 20 variables (input parameters) but only 100 "backtest trades" even the best equity curve is useless resp. can't tell you anything about your system's real profitability. However when having e.g. 6 (core) variables and 4000 (backtest) trades with a smooth equity curve chances are good to have developed a stable strategy. That's all a matter of statistics and their evaluation. In your case everything is fine apart the fact that you cant' conduct more "test trades". That's not your fault, it's simply an issue due to insufficient price data. In any case I wish you good luck and that your forward test confirms the backtests.
Thank you.
Why the Steed-Signal is deleted?
Why the Steed-Signal is deleted?
Hi!
The signal will be available next week. Some technical problems.