So i think i am going back to NT (after using eSignal for a while) to do some strategy coding and i just wanted to get some advice before i start regarding multi timeframe and multi instrument implementation.

I previously found that NT was giving me some problems with multi time frame and instrument, sometimes returning incorrect values (perhaps bc my code was not correct). I understand the process of calling another bars_index and the general mechanics for how it works in NT.

My question is for those users who have coded indicators and more specifically strategies using multi time frame and multi indicators on NT 7 what has been your experience. I read through most of the older posts on the topic and it seems like you can get multi time frame and multi instrument to work just fine

I'm just want to make sure that is the case before i reinvest my time again to get back into the NT coding realm

Thanks for your input

Let me just mention some of the critical points:

(1) An incoming tick usually triggers the primary bar series (BarsInProgress == 0) first, the secondary series (BarsInPrgoress ==1) is triggered afterwards. If you have fixed period bars such as minute bars, they may have an identical close. In this case the values calculated from the secondary bar series will not be available, when the primary bar series closes. The MTF indicator will display a lag.

The indicators that I have coded have an option to reaffect the calculated value to the primary bars with the first incoming tick of the next primary bar. This is a one-tick repainting, which may eliminate the lag for discretionary traders.

(2) If you do not want that your MTF indicator displays as a step function, you need to interpolate the values calculated vai the secondary bars. You will need to use different algorithms for interpolation for equidistant and non-equidistant bar spacing.

(3) The logic for affecting values to the primary bars, can be slightly different on historical (1-tick) bars compared to real-time bars. You need to take into account the specific behavior of NinjaTrader at session breaks as well.

(4) The adding of daily data series does not work correctly, if you use session templates with multiple sections, such as the RTH and ETH sections.

I think that's what I have seen so far.


Otherwise the multi-timeframe indicators work well and can be applied for two specific cases:

(A) Calculate higher timeframe values to be used as trendfilters

(B) Achieve higher accuracy for indicators like VWAP, Market Profile or Opening Range by calculating the values from higher resolution bar series

Thanks Fat Tails. Your posts have been really insightful

Yeah the problem with daily data I think is where I was having problems before. I like looking at the normal and overnight session for data. I was thinking of just reading it from a text file.

I am really only interested in historical backtesting for now. I would like to use 1min data as the main data series but have higher time frame data in the conditions for my strategy (ie if the 1min low is less than low(3) using daily data do something or, if the low(0) using 1min data is less than the 30min moving average .,.. etc.

Do you know if this would work easily?

Its a great point about identical closes on the bar series and the lag. I didnt realize there was a lag, but it makes perfect sense. I guess it shouldnt be too hard to work around

Also are you aware of any example of code that I can look through for the more advanced stuff

eurostoxx 

Thanks Fat Tails. Your posts have been really insightful Yeah the problem with daily data I think is where I was having problems before. I like looking at the normal and overnight session for data. I was thinking of just reading it from a text file. I am really only interested in historical backtesting for now. I would like to use 1min data as the main data series but have higher time frame data in the conditions for my strategy (ie if the 1min low is less than low(3) using daily data do something or, if the low(0) using 1min data is less than the 30min moving average .,.. etc. Do you know if this would work easily? Its a great point about identical closes on the bar series and the lag. I didnt realize there was a lag, but it makes perfect sense. I guess it shouldnt be too hard to work around Also are you aware of any example of code that I can look through for the more advanced stuff

I have not yet tested multi-timeframe strategies, but I am sure that it requires some skill. I am still trying to figure it out myself.