Why sell profitable EAs! - page 8

New comment
 

Dear Mathemat,

can you say a few words about the system from the backtest above ?
 
Somehow I don't understand how this is possible.... Is it due to a large number of mismatches?
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Period 1 Hour (H1) 1998.11.06 11:00 - 2008.03.19 08:00 (1989.01.01 - 2009.03.12)
Model All ticks (most accurate method based on all smallest available timeframes)
Bars in history 56154 Modelled ticks 11503347 Modeling quality n/a
 
timbo:
Somehow I don't understand how this is possible.... Is it due to a large number of mismatches?
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Period 1 Hour (H1) 1998.11.06 11:00 - 2008.03.19 08:00 (1989.01.01 - 2009.03.12)
Model All ticks (most accurate method based on all smallest available timeframes)
Bars in history 56154 Modelled ticks 11503347 Modeling quality n/a

Corrected .There's a number eating away at it.
 
timbo:
Somehow I don't understand how this is possible.... Is it due to a large number of mismatches?
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Period 1 Hour (H1) 1998.11.06 11:00 - 2008.03.19 08:00 (1989.01.01 - 2009.03.12)
Model All ticks (most accurate method based on all smallest available timeframes)
Bars in history 56154 Modelled ticks 11503347 Modeling quality n/a

Yes, when the number of mismatch errors is too high, there is simply no assessment of modelling quality.
 
Rosh:
Yes, when the number of mismatch errors is too high, there is simply no assessment of modelling quality.
I.e. this report below the n/a line is not even readable, because there is no point. Thank you.
 
Igonter: In the case of pips/scalping it is a question of the quality of quotes (filtering effects, running on quotes from another source).
This is a bit more complicated. Traders do not really know what kind of filtering is used in brokerage companies. Well, stop levelling should be changed as it was in the case with vinvin, frizzlevel - what else? The filters that restrict pipsing can be changed by the brokerage company at any moment - even without formal change of parameters of the MarketInfo() function. And testing of pipsewing strategies on sources other than the one we are going to use, in my opinion, does not make much sense. Guarantees are very bad here...
 
Rosh:
timbo:
Somehow I don't understand how this is possible... Is it due to a large number of mismatches?
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Period 1 Hour (H1) 1998.11.06 11:00 - 2008.03.19 08:00 (1989.01.01 - 2009.03.12)
Model All ticks (most accurate method based on all smallest available timeframes)
Bars in history 56154 Modelled ticks 11503347 Modeling quality n/a

Yes, when the number of mismatch errors is too high, there is simply no assessment of modelling quality.

How can you help or what advice can you give?
 
azfaraon:

Dear Mathemat,

can you say a few words about the system from the backtest above ?
I am not a matamat, but I will say about the system that it is a total sax. 300% in 10 years ?
Buy yourself some bonds: for 10 years practically the same money, but almost no risk and certainly no hassle with DCs
 
Mathemat:
Igonter: in the case of pips/scalping, the issue is the quality of the quotes (effect of filtering, running on quotes from another source).
It will be more complicated here. Traders do not really know what kind of filtering is used in brokerage companies. Well, stop levelling should be changed as it was in the case with vinvin, frizzlevel - what else? The filters that restrict pipsing can be changed by the brokerage company at any moment - even without formal change of parameters of the MarketInfo() function. And testing of pipsewing strategies on sources other than the one we are going to use, in my opinion, does not make much sense. Guarantees are very bad here...

You don't need to know how the filter is set up or if it is there. We just need to know if the strategy works with some "benchmark" quotes from an average brokerage company or not. For scalping, it is important, because many small brokerage companies intentionally disable the filter to attract people. But we will not take a big depo there and even if we do take it back, we will not get it back. :)
 

azfaraon, if there were not so many chart mismatch errors and modeling quality was about 90%, and if the report corresponds to playing with 0.1 lot, then first impression - not bad at all (about 2000 pips per year, i.e. about 170 per month).

If the real game will be with a different MM (most likely it will be), then be sure to make a report and for such MM, so that then it does not get bad.

I noticed a chip: 1 Hour (H1) 1998.11.06 11:00 - 2008.03.19 08:00 (1989.01.01 - 2009.03.12). Curious...

The tip to correct the mismatch errors is standard: delete all history cables in hst format on all TFs in your brokerage company folder and download history from MQ from the History Center (if you don't want to overwrite your brokerage company quotes, you better do it all on a test copy of the terminal). The second option: use the period_converter script directly, if you think your timeframe is good enough. But it would still be good to erase the history on larger TFs.

P.S. These words should not be regarded as a guide for immediate actions in trading on the system. A normal, complete testing has not been canceled. And one more thing: it is better to make the initial capital the same as the one you are going to bet later.

1234567891011
New comment
Reason: