MultiCharts: Trading Software for Automated Trading and Backtesting

I have read the wiki section on chart resolution and how checking the Format Instrument > Settings - "Build from Minutes" for 1 Day charts relates to sessions.

Just to be sure, should I check the build from minutes for Daily resolution when using Oanda Forex?

Thank you.

I have read the wiki section on chart resolution and how checking the Format Instrument > Settings - "Build from Minutes" for 1 Day charts relates to sessions.

Just to be sure, should I check the build from minutes for Daily resolution when using Oanda Forex?

Thank you.

"Build from Minutes" is used because...

You do not want to use the official "Daily" session data provided by your broker.
You want to define a custom daily session, and you want to build your "Daily" bars only using data from the custom session.

eg. a custom daily session with the starting time of 11:00 am and ending time of 6:00 pm.

Thank you TJ,

That makes sense now. So on the Format Instrument > Settings Tab under Chart Settings I see the Sessions field. I have it set to Default, maybe this is not correct or does it no that Forex is 24/5? In any event, using the same setting under Sessions produces different results when checking Build from Minutes, yet there is no custom session created. How exactly do you create a custom session. I just see the drop down menu with lots of preset options.

What is the Default session?
Should I use 24/5 for Forex if I don't want a custom session?
Whichever session is selected it will apply to the live chart, back testing and performance reports. Correct?

I really appreciate all your help. There are so many nuances with MC's back testing to ensure accuracy. Accuracy meaning that the settings that are applied to back testing will be the same as you would expect to get in real trading. I learned some of this the hard way by not using Bar Magnification. Which I am still unsure if it is needed if if entry/exits are based on bar close/open. I understand that it is needed for accuracy for intra-bar activity like stops or take profit areas. Teh same for IntraBarPersist.

John

Hi John:

MultiCharts is a very powerful and flexible program. ie. it has lots of features and optional parameters. It will take some time for a new user to learn all the detail. So take your time and enjoy.


For sessions,
please see post #4
viewtopic.php?f=16&t=6845

For backtesting, start here...

Autotrade / Backtest / Optimization FAQ
viewtopic.php?f=16&t=10811

TJ,

Thanks. I have read most of this. Not everything is exactly clear though. It should be a simple as "if you don't use intrabar data i.e. SL/TP/Limit orders and only use Open and Close values then you do not need the bar magnifier or intrabar features."

MC support tells me that setting the Bar Magnifier to ticks is the most accurate. It makes sense if you are using set SL/TP's because as the documentation states, the backtester does not know if the stop loss or take profit was hit first, it only knows that those rates were within the OHLC of that bar. Of course this could be very inaccurate. Support then tells me in the same response that if I am using Daily chart resolution that if I use the tick precision that my results are most likely unrealistic.

See the confusion. Next step is to manually test data activity to see if the calculations work correctly. Any advice here would be helpful. Maybe the playback? Signing off...Good night and thank you again.

John

John

TJ,

Thanks. I have read most of this. Not everything is exactly clear though. It should be a simple as "if you don't use intrabar data i.e. SL/TP/Limit orders and only use Open and Close values then you do not need the bar magnifier or intrabar features."

MC support tells me that setting the Bar Magnifier to ticks is the most accurate. It makes sense if you are using set SL/TP's because as the documentation states, the backtester does not know if the stop loss or take profit was hit first, it only knows that those rates were within the OHLC of that bar. Of course this could be very inaccurate. Support then tells me in the same response that if I am using Daily chart resolution that if I use the tick precision that my results are most likely unrealistic.

See the confusion. Next step is to manually test data activity to see if the calculations work correctly. Any advice here would be helpful. Maybe the playback? Signing off...Good night and thank you again.

John

John

You are expanding your scope... the scenario is no longer simply "buy next bar open" and "sell this bar close". You are adding facades to the Rubic's cube. Your complexity has just been increased exponentially.


:v

Next step is to manually test data activity to see if the calculations work correctly. Any advice here would be helpful. Maybe the playback? Signing off...Good night and thank you again.

Data playback indeed is an option, but keep in mind that the playback speed is set by the user (e.g., 3 updates per second), which might not be realistic if the instrument you're "data playbacking" traded volatile. In that case there would be a lot more updates than 3 per second. So in other words, data playback doesn't give an accurate picture of a strategy's slippage.

That said, confirming calculations with the data playback is of course possible. But for testing your strategy (before trading) a test with real-time, accurate data with a demo account from your broker is probably your best bet.

By the way, using a daily chart with the Bar Magnifier set to tick precision seems overkill to me. It's indeed the most accurate, but minute data will also suffice. That's because I think a daily strategy shouldn't be that sensitive to intra-day ticks (after all, the point of a daily strategy is to eliminate part of the noise - right?).

Another drawback of using the most accurate backtesting setting it's that its easy to fool yourself into thinking that the results are an accurate reflection of how the strategy would have performed. But that's my unsolicited view on it.

Josh,

Very useful and clear explanations. Finally. Thank you.


John