How can I make 30m, 1H, 1Day historical data from 1m data I have?

New comment
 

Hi, 

I have found good historical data to import for backtesting and it's in 1 min timeframe.

I can import it just fine.

However, I don't have any clue how I test for 15 min, 1hr and 1 day.

It should be pretty simple to derive all the other time frames from 1min.

What's the correct procedure to do it?

Thanks,

SilverPrince

 
SilverPrince:

Hi,

I have found good historical data to import for backtesting and it's in 1 min timeframe.

I can import it just fine.

However, I don't have any clue how I test for 15 min, 1hr and 1 day.

It should be pretty simple to derive all the other time frames from 1min.

What's the correct procedure to do it?

Thanks,

SilverPrince

There is a script in your MT4 called period_convertor. It calculates the other timeframes but you need to be careful for the filename of your history.

 

Hi SilverPrince,

can you let me know where to get the good historical data from? I have been looking for it for a long time already,

thanks

 

I have got the datasource in here:

http://www.forextester.com/datasources.html

And currently working through period_converter to convert 6 major pairs to all time frames.

I've seen somewhere tick data too.

But you have to manually download them for every hour. So unless some perl programmer writes a parsing HTML script - it's really tough to get all ticks even for a month.


Alternatively, I just found out that Meta Trader has a bunch of brokers with demo accounts built-in - so maybe I will try to get demo acct with all of them and download Historical Data from all of them

and then keep a backup.


The important thing is to keep your backtesting MetaTrader OFFLINE - so the automatic download does not screw up your historical data.


I'll keep you posted what my findings are.

 

thanks a lot SilverPrince. I will try them. i have another problem. after i downloaded these historical data and used them,now my computer got frozen often.

 

I have download that data and converted it to M5 and H4 what I need. but the model quality is too low (25%), so I still need looking for another better data.

thanks anyway.

 
MQL_MM:

I have download that data and converted it to M5 and H4 what I need. but the model quality is too low (25%), so I still need looking for another better data.

thanks anyway.

If you run M1 data in the backtester 25% is the best you can get regardless who supplies the data.

 

I opened demo accounts and downloaded historical data from 3 different brokers for all major currency pairs.

The test results on all 3 of them are quite similar (providing you use the same leverage for all of them). Broker data differs for 1 or 2 pips - so in several years you get 10-20-30 dollars of difference.

That's quite all right - at least I get consistent results


I also successfully converted ForexTester data using Period_converter and period_converter_optimized scripts.

However, the test results are MAJORLY different from the MetaTrader supplied data sources (from the 3 brokers above).

I tried different time periods - same thing - very different results on ForexTester data.

Also, when I test on the ForexTester data source for roughly the same number of bars I get much much fewer ticks simulated for some reason.

That's quite scary :(


So, unless period_converter is broken... at this point of time, I'd say the simulation is pretty useless.

And without real tick data - there is almost no point in testing. ..... :(


or.. Unless somebody who does have real tick data steps up and does test for both real tick data and simulated data from say Alpari and tells us if there is much difference on several EAs...

New comment
Reason: