Backtesting/Optimization - page 75

[Deleted]  

question

hello guys i m not new to this forum i read a lot and post really less i would appriciate any kind of help thanx in advance

so let s came to it :

i ve got an ea that on the backtesting is doing the job as i wish but when i put it on a demo in live it does not what i want so

1 it goes only short ( the code is the same for short and long )

2 in backtesting with the simulator it works fine

why ?

again thanx for any help

thomada
3
thomada  

EA testing

Hi guys,

I am a beginner in automatic trading (in trading in general) and using MetaTrader 4. I have started to build an EA and after hard time finished to implement it. I did some backtest with it and I have got some strange result:

the chart is growing but after a while falling. When I look into the report it is corresponding to "close at stop". What does it mean ?

If my EA is a good one and if you help me to improve it, I will share it.

Thank you for your help

Files:
Guyver
1152
Guyver  
thomada:
Hi guys,

I am a beginner in automatic trading (in trading in general) and using MetaTrader 4. I have started to build an EA and after hard time finished to implement it. I did some backtest with it and I have got some strange result:

the chart is growing but after a while falling. When I look into the report it is corresponding to "close at stop". What does it mean ?

If my EA is a good one and if you help me to improve it, I will share it.

Thank you for your help

Well, what is happening is that "Strategy Tester" reached the end of the test period but carried a position which was still open. Now since there is no further history available the tester forces the position to close that is all. no biggie

-guyver

robinhood100
20
robinhood100  

Hi,

I use strategy generator (Forex Strategy Generator - freeware) and see it saves a LOT of time to check different strategies.

Mainly I use 5 min and 15 minutes to see if the strategies are working at all.

Which philosophy is the best to create the perfect strategy:

1. Using only one specific time frame and currency pair (for example EUR/USD and 5 Min):

a. Use only 1 week/1 month of market data to generate strategy (after all it is current state of market)

b. Use few years of data on the specific time frame

2. Trying to find strategy which will be getting profits on all time frames in given currency pair

a. Use only 1 week/1 month of market data to generate strategy (after all it is current state of market)

b. Use few years of data on the specific time frame

3. Trying to find strategy which will be getting profits on few main currencies (EUR/USD, GBP/USD, USD/JPY, EUR/GBP) in one time frame

a. Use only 1 week/1 month of market data to generate strategy (after all it is current state of market)

b. Use few years of data on the specific time frame

4. Trying to find only one time frame which is most reliable and most noise free (longer time frames like 1h, 4h, 1d)

a. trying to find the strategy working on all main currencies (EUR/USD, GBP/USD, USD/JPY, EUR/GBP) in the same time frame

Please tell what are your experiences with forward testing strategies and which above philosophy is the best?

Guyver
1152
Guyver  
robinhood100:
Hi,

I use strategy generator (Forex Strategy Generator - freeware) and see it saves a LOT of time to check different strategies.

Mainly I use 5 min and 15 minutes to see if the strategies are working at all.

Which philosophy is the best to create the perfect strategy:

1. Using only one specific time frame and currency pair (for example EUR/USD and 5 Min):

a. Use only 1 week/1 month of market data to generate strategy (after all it is current state of market)

b. Use few years of data on the specific time frame

2. Trying to find strategy which will be getting profits on all time frames in given currency pair

a. Use only 1 week/1 month of market data to generate strategy (after all it is current state of market)

b. Use few years of data on the specific time frame

3. Trying to find strategy which will be getting profits on few main currencies (EUR/USD, GBP/USD, USD/JPY, EUR/GBP) in one time frame

a. Use only 1 week/1 month of market data to generate strategy (after all it is current state of market)

b. Use few years of data on the specific time frame

4. Trying to find only one time frame which is most reliable and most noise free (longer time frames like 1h, 4h, 1d)

a. trying to find the strategy working on all main currencies (EUR/USD, GBP/USD, USD/JPY, EUR/GBP) in the same time frame

Please tell what are your experiences with forward testing strategies and which above philosophy is the best?

I am sure everyone will have a different view . When i started getting serious i wanted to scalp EUR and nothing else and i am stick to it till date . so i looked/create/re-create/trashed/un-trashed ideas on a 2-3 years of data and then update my strategies as market changed. Scalping using Fibonacci S/R or scalping 15min EMA taking anything from 6 to 10 pips and occasionally 30 pips is what has worked for me ( manual/automated ). I am sure everyone has their own style as well.

below is just as example of 2 of the recent trades today ( yesterday now ).

sunshineh
1366
sunshineh  

Backtest Data

Hi,

I tried a backtest in H4:

I have on my Alpari Account a modeling quality of 90% but only one green colour on the report.

Than I have an additional Account with ist running for a few month, there I have a modeling quality of 62,4% as you can see on my attachment. There are more colours on the report.

Which report is better?!

I heard, that it is good to have many colours?

What does it mean?

Thank you!

Files:
Guyver
1152
Guyver  
sunshineh:
Hi,

I tried a backtest in H4:

I have on my Alpari Account a modeling quality of 90% but only one green colour on the report.

Than I have an additional Account with ist running for a few month, there I have a modeling quality of 62,4% as you can see on my attachment. There are more colours on the report.

Which report is better?!

I heard, that it is good to have many colours?

What does it mean?

Thank you!

To keep it short

When you start a backtest like in 4 hrs.. it will take data from 1 minute upto 4 hrs that is all the time frames which makes the 4hr bar ( Meta trader )...

now.. the lime color shows where all the data from 1 minute to 4hr was available .. all other time frames where some data was missing or in complete ( including limitation by date ( Grey ) ) etc.

i think you get the point ..

Mike
237
Mike  

Chain Optimization (Strategy Tester)

Hi guys,

I have a question/idea...

Optimizing variables in metatrader's strategy tester is a double edged sword. Its hard to tell whether a successful result succeeds because the idea actually works, or whether its a statistical result from the many failed variations on the same variables.

So.. what I was wondering is this.... wouldn't this conundrum be eliminated if one were to optimize in such a manner:

1. Optimize all variables for a 3 month period. Pick the best result.

2. Test the optimization on 1 day period immediately following the 3 month period.

3. Repeat over and over, each time moving the time frame that you're optimizing for by 1 day.

4. If the cumulative 1-day trades are significantly positive, that is truly indicative of a working strategy.

rjay
58
rjay  

"exchange rate cannot be calculated"

I'm trying to backtest some EA's against Futures (_DJI, etc.) at Alpari but am not getting any trades because of these two errors:

2010.07.05 14:17:57 Tester: exchange rate cannot be calculated

2010.07.05 14:17:57 Tester: margin exchange rate cannot be calculated

Anyone know how I fix this ?

zipfrog
110
zipfrog  
rjay:
I'm trying to backtest some EA's against Futures (_DJI, etc.) at Alpari but am not getting any trades because of these two errors:

2010.07.05 14:17:57 Tester: exchange rate cannot be calculated

2010.07.05 14:17:57 Tester: margin exchange rate cannot be calculated

Anyone know how I fix this ?

That error is because the EA's are only designed for currencies. You cannot use them for Dow futures etc...

Regards,

Zipfrog