External data vs Broker's own historical data for backtesting ?
Hi traders and coder.
I have a dilema of choosing between external data and broker's own historical data.
The reason behind this is that broker's own historical data only provide me 30% modelling qualilty plus broker's own histrical data only goes back to 1 year.
By downloading historial data from histrocal center in MT4, I am sure I can increase the modlling quality much higher, something like 99%.
However, for live trading, I am really reluctant to use external data rather than broker's own historical data.
The reason for this is that our mechanical entry and exit generated from our trading strategy should be best fit to the data where broker will execute trading order (This claim may not be relevant if broker and we are looking at different data. However, I assume broker's provide data with high integirity. :) ).
Which one would recommend ? Will you still recommed to use external data with 99% modelling quality ? On the other hand, the poor modellng quality of broker's own data might means that data itself not consistent among different time frame.
This is not easy to decide what to do. Sharing of your experience will be really apprecaited.
Kind regards.
No one should provide settings with EA, it should be optimized for the user's platform, user's broker, user's computer. Concerning modelling quality, everybody download datas from MT to get the best quality as MT scores it. Is that to mean that any others' data aren't good ? I too assume brokers' data reliability.
EDIT : MTs' coders seems to be aware of as MT5 came without such a feature.
No one should provide settings with EA, it should be optimized for the user's platform, user's broker, user's computer. Concerning modelling quality, everybody download datas from MT to get the best quality as MT scores it. Is that to mean that any others' data aren't good ? I too assume brokers' data reliability.
EDIT : MTs' coders seems to be aware of as MT5 came without such a feature.
Good things about MT5 strategy tester is that they give you quite consistent result.
However, the results from EA differ quite a lot for MT4 strategy tester especially when you use broker's own data. It seems the results become much consistent when I use data downloaded from Meta Quote.
However, this data is not necessarily the data we are looking at through broker's MT4 charts. This is model building irony keep coming back to me many years.
Not sure how to cope with this data issue yet.
Optomize on MT data and test on broker's one.
Your approach may work and probably good thing to do. However, we will only catch some good results out of all optimization efforts. Rest of optimization results may be waste due to data mismatch between your historical data and your broker's data.
One more thing to ask is how do you switch back to your broker's data after you have downloaded historical data Meta Quote using history centre except reinstalling another Meta Trader platform from the same brokers? Is there such a way ?
Good things about MT5 strategy tester is that they give you quite consistent result.
However, the results from EA differ quite a lot for MT4 strategy tester especially when you use broker's own data. It seems the results become much consistent when I use data downloaded from Meta Quote.
However, this data is not necessarily the data we are looking at through broker's MT4 charts. This is model building irony keep coming back to me many years.
Not sure how to cope with this data issue yet.
What I mean by inconsistent results is that when I use broker's data and if I do optimization on open price with their data, I don't get the same results like profit and drawdown, etc when I do backtesting with one of the parameters found from optimization.
Sometimes profitable setting in optimization turns to negative loss in single back testing.
However when I download data from Meta Quote using history centre, the results of parameter setting from optimization seems similar to single back testing.
So I can say that the modelling quality plays an important role. However modelling quality does not solve the problem of discrepancy between the Meta Quote historical data and broker's data.
Kind regards.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hi traders and coder.
I have a dilema of choosing between external data and broker's own historical data.
The reason behind this is that broker's own historical data only provide me 30% modelling qualilty plus broker's own histrical data only goes back to 1 year.
By downloading historial data from histrocal center in MT4, I am sure I can increase the modlling quality much higher, something like 99%.
However, for live trading, I am really reluctant to use external data rather than broker's own historical data.
The reason for this is that our mechanical entry and exit generated from our trading strategy should be best fit to the data where broker will execute trading order (This claim may not be relevant if broker and we are looking at different data. However, I assume broker's provide data with high integirity. :) ).
Which one would recommend ? Will you still recommed to use external data with 99% modelling quality ? On the other hand, the poor modellng quality of broker's own data might means that data itself not consistent among different time frame.
This is not easy to decide what to do. Sharing of your experience will be really apprecaited.
Kind regards.