hello and good day to everyone,
i have developed a number of automated strategies both on easylanguage and ninjascript. i have a modus operandi that i haven't been able to follow on ninjatrader, neither for back-testing - optimizing strategies over historical data nor to have strategies place trades in real time on a simulated account. that's why i would like to read opinions and suggestions from traders who have experience executing automated strategies live on ninjatrader.
my first issue with ninjatrader is that the platform does not include the kase bar chart type. a number of people, including myself, have requested on this forum that the platform include this feature but so far ninjatrader has refused to honor these requests.
the following is a description of the modus operandi i typically use to evaluate how good any automated strategy is, and how i haven't been able to do this on ninjatrader. i have already emailed the people at ninjatrader's support department but we haven't been able to pinpoint where the problems could be, that's why i think experienced traders and programmers could know better.
in other platforms, i normally load historical data for an instrument from 01-I-2006 or 01-I-2009 to 31-XII-2013 to a chart and optimize my strategies over that period, leaving all the data from 2014 and later out of sample. then i just add the information that had been left out to the chart, so that it would go from 2006 or 2009 and up to date. with this methodology, all the trades after 2013 to this day have never been optimized and can be considered to be a good approximation of how an strategy would do if trading live.
however, with ninjatrader, this method doesn't work anywhere near the way it should. first, when i create a chart, i get the impression that historical data will only go as far back as 2010 for any instrument. i can't tell for sure, as i don't know how to tell what date does the first bar in a chart correspond to. after i optimize the strategy's parameters in a strategy analyzer window, no matter if it is from 2006 or 2009 to 2013 or up to date, the results i get are always aberrant. there is only a very small number of trades and they all correspond to just one single month in one single year, when i really would be expecting trades almost every single day over a period of 8 to 10 years. i have tried to optimize several different strategies i have developed myself on different instruments over long time periods and this malfunction seems to always happen.
so far i have only used cqg's free trial data, i don't know if this could be the problem if there were limitations on the historical data that is available on nt. also, i have even thought of going through the hassle of translating my strategies from nt7 to nt8 as i don't know whether this could help in case the nt8 platform was much better. are these issues i have mentioned normal? does market data and the ninjatrader platform operate only in a limited capacity under the free trial?
ninjascript is far more powerful and allows for far more detailed coding than easylanguage does, so if trading results were comparable i would definitely prefer to use ninjatrader. even more crucially, ninjatrader's total costs are significantly lower (even more so for eurex products), so once again, if results were comparable i would definitely trade with ninjatrader, but as i will prove with some experiments i have carried out, so far i have only had results that are far from being comparable even for the same strategy, over the same instrument, using the same bar type and size, the same slippage, the same commissions, the same everything.
i would appreciate any suggestions or recommendations from experienced traders and programmers about how to make sure i can get all the historical data from 2006 to 2013 for the instruments i want to evaluate be able to evaluate my strategies over and whether what i typically do can in fact be done work on ninjatrader. later i will be sharing some comparable back-tests and optimizations for the sample 2 moving average crosses strategy i have carried out on the fdax, nq on different bar types that are radically different on ninjatrader compared to other platforms to see how these discrepancies could be solved.
thanks a lot, regards.
i have developed a number of automated strategies both on easylanguage and ninjascript. i have a modus operandi that i haven't been able to follow on ninjatrader, neither for back-testing - optimizing strategies over historical data nor to have strategies place trades in real time on a simulated account. that's why i would like to read opinions and suggestions from traders who have experience executing automated strategies live on ninjatrader.
my first issue with ninjatrader is that the platform does not include the kase bar chart type. a number of people, including myself, have requested on this forum that the platform include this feature but so far ninjatrader has refused to honor these requests.
the following is a description of the modus operandi i typically use to evaluate how good any automated strategy is, and how i haven't been able to do this on ninjatrader. i have already emailed the people at ninjatrader's support department but we haven't been able to pinpoint where the problems could be, that's why i think experienced traders and programmers could know better.
in other platforms, i normally load historical data for an instrument from 01-I-2006 or 01-I-2009 to 31-XII-2013 to a chart and optimize my strategies over that period, leaving all the data from 2014 and later out of sample. then i just add the information that had been left out to the chart, so that it would go from 2006 or 2009 and up to date. with this methodology, all the trades after 2013 to this day have never been optimized and can be considered to be a good approximation of how an strategy would do if trading live.
however, with ninjatrader, this method doesn't work anywhere near the way it should. first, when i create a chart, i get the impression that historical data will only go as far back as 2010 for any instrument. i can't tell for sure, as i don't know how to tell what date does the first bar in a chart correspond to. after i optimize the strategy's parameters in a strategy analyzer window, no matter if it is from 2006 or 2009 to 2013 or up to date, the results i get are always aberrant. there is only a very small number of trades and they all correspond to just one single month in one single year, when i really would be expecting trades almost every single day over a period of 8 to 10 years. i have tried to optimize several different strategies i have developed myself on different instruments over long time periods and this malfunction seems to always happen.
so far i have only used cqg's free trial data, i don't know if this could be the problem if there were limitations on the historical data that is available on nt. also, i have even thought of going through the hassle of translating my strategies from nt7 to nt8 as i don't know whether this could help in case the nt8 platform was much better. are these issues i have mentioned normal? does market data and the ninjatrader platform operate only in a limited capacity under the free trial?
ninjascript is far more powerful and allows for far more detailed coding than easylanguage does, so if trading results were comparable i would definitely prefer to use ninjatrader. even more crucially, ninjatrader's total costs are significantly lower (even more so for eurex products), so once again, if results were comparable i would definitely trade with ninjatrader, but as i will prove with some experiments i have carried out, so far i have only had results that are far from being comparable even for the same strategy, over the same instrument, using the same bar type and size, the same slippage, the same commissions, the same everything.
i would appreciate any suggestions or recommendations from experienced traders and programmers about how to make sure i can get all the historical data from 2006 to 2013 for the instruments i want to evaluate be able to evaluate my strategies over and whether what i typically do can in fact be done work on ninjatrader. later i will be sharing some comparable back-tests and optimizations for the sample 2 moving average crosses strategy i have carried out on the fdax, nq on different bar types that are radically different on ninjatrader compared to other platforms to see how these discrepancies could be solved.
thanks a lot, regards.
Comment