indicator lag at open

[ATI user]

I have yet to get the same results running my cs in Strategy Analyzer that I get running it in Market Replay. This is unfortunate due to the huge time difference running MR.

The reason for this is indicator lag when SA starts.

Lets say you want to test only one day 3 nov. If your session time is 12:00AM to 12:00AM and you run SampleMACrossover or something similar (ie. with very basic indicators) then there is no lag and you get good results.

However, if you use an indicator in your cs like ADXVMA, the lag is so huge that the results are useless. Even though I use the 1200AM start time and my cs does not trade until 930AM, the ADXVMA does not converge to useful price levels until after noon.

3 nov is a monday, so I even tried to backtest from sunday 2 nov through 3 nov and still got useless results.

is there a way to have SA consider ALL pre-market data (i.e. prior days) for the sake of the indicators?

in MR of course, as long as you have replay data prior to the test date, your indicators are bang on.

thanks.

[NinjaTrader_Jason]

I am not sure how historical data the ADXVMA indicator needs to calculate its values. I suggest to ensure the required historical data is requested upon performing the backtest.

You will need to restrict your strategy so it will not submit orders for the period of historical data that is needed to calculate the indicator values.

[ATI user]

I am not sure how historical data the ADXVMA indicator needs to calculate its values. I suggest to ensure the required historical data is requested upon performing the backtest.

my point is that SA should have the same lookback capability as MR so you can get the exact same results using both. SA is limited in this regard.

You will need to restrict your strategy so it will not submit orders for the period of historical data that is needed to calculate the indicator values.

although my cs will start trading at any time during the session, that would be pointless. I should not have to change my trading time to make SA work. SA should work for the whole day, from the Open, and would if it had a better lookback to populate ALL indicators with data

[NinjaTrader_Jason]

I will forward your suggestion to our development team and ask them if they can add this to the list of future considerations for the software.

Thank you for your feedback.

[ATI user]

duplicate post

[ATI user]

the difference appears to be that SA and MR use two different data sources

if you run a simple MA crossover on a 100 tick chart with a 100 period EMA, the results are drastically different (if you did not record pre-market data from the prior day's close).

MR appears to use whatever data you have recorded while SA gets its data from somewhere else....NT servers I assume

therefore it will always be impossible to get the same results from SA and MR...unless you happen to record the exact same data in MR...unlikely unless you leave you computer on 24 hours a day to record?

question 1: what is the source of SA data so I can try to approximate results?

question 2: why are the sources not the same? i.e. if SA can pick up a complete data set from NT servers, why do we have to record data for MR?

question 3: does NT not agree that you should be able to get the exact same results from running the same settings in both SA and MR? if not, why not? i.e. what is the point of having 2 different results?...they have to be the same or one is useless isn't it?

[NinjaTrader_Dierk]

Market replay gets data from the market replay data pool whereas backtest is triggered of the historical chart data which is different.

[ATI user]

question 2: why are the sources not the same? i.e. if SA can pick up a complete data set from NT servers, why do we have to record data for MR?

question 3: does NT not agree that you should be able to get the exact same results from running the same settings in both SA and MR? if not, why not? i.e. what is the point of having 2 different results?...they have to be the same or one is useless isn't it?

[NinjaTrader_Dierk]

Sim account fills work an MR as on a live feed which is very different than fills on backtest.

[ATI user]

Sim account fills work an MR as on a live feed which is very different than fills on backtest.

so SA is only historical (closed) bars while MR is tick by tick (intra-bar) data?...which is why one would want to (need to) use MR for a cs that makes decisions based on tick by tick data instead of closed bars ....right?

in that case, my cs, which uses CalculateOnBarClose = false, will not get accurate results in SA ever...correct?

[NinjaTrader_Dierk]

>> so SA is only historical (closed) bars while MR is tick by tick (intra-bar) data?
Correct. MR fills even are based on bid/ask which is not available historically.

[ATI user]

Ok ...thanks Dierk

so to take advantage of SA I will put in a switch to only use closed bar data...

this, unfortunately, does not resolve the indicator lag on the open problem...any thoughts on having SA look further back than the session open of the current day being testing to get data to populate the slower indicators? ...I believe this is needed...even for such basic indicators as a 100 period EMA on a 420 tick chart

[NinjaTrader_Dierk]

>> any thoughts on having SA look further back than the session open of the current day being testing
Unfortunately this is not supported.