Hi Everybody,

I am trying to figure out why when I run the historical tester there is a slight difference between the parameters of my strategy and the graphical plot of it.

In my case, I have setted a buy/sell order when the 1H bar cross the SMA(200). As you can see in the attached file the order, according to my strategy, should have been filled on the red-circled candle but it is filled some candles afterwards instead...

Can't understand where the problem is...

These are my parameters:

instrument = Instrument.EURUSD;
selectedPeriod = Period.ONE_HOUR;
smaTimePeriod = 200;
indicatorFilter = Filter.WEEKENDS; ( I have tried to change this one but it doesn't change anything at all)
candlesBefore = 200;
candlesAfter = 0;

double sma = indicators.sma(instrument,selectedPeriod,OfferSide.BID,AppliedPrice.CLOSE,smaTimePeriod,indicatorFilter,candlesBefore,prevBar.getTime(),candlesAfter)[0];

Thank You for your Help!

For appropriate comparsion, you should use the same filter both on chart and strategy.
Consider logging the relevant values from your strategy (i.e. indicator outputs, bar times, prices, etc.) to see what execution paths get taken and why, see:
https://www.dukascopy.com/wiki/#IConsole/Logging_values

Thanks, I added this:

console.getOut().println("Instrument:" + instrument + "sma:" + df.format(sma));

but I can't find out where I should see it, it doesn't appear in the final report, and no exception has been thrown therefore there shouldn't be any error.

"slight difference" - this bit made me laugh - the space there is big enough to park two cars

ok so here's the thing. either:

a) your filtering is off such that what you think your strategy is seeing in terms of price or SMA at time t isn't what you think it is
b) the piece of code you are using to determine the cross is broken
c) the essential elements are not written to deal with the testing scheme (1H, 4OHLC ticks) you've selected
d) two but not three of the above are true
e) three of the first three are true

the question is, are you just telling us your woes just so we know, or are you telling us so we can really try to help you?

No instrument subscription / feed descriptor logic you might use
No look at the crossover logic
No indication what you are doing with that 'double sma' you pulled with [0] which is for the *'previous candle' (edit: *not true - see below)
No indication as to whether your observation is one-off or noticed on every entry - and further if the observation follows any
pattern e.g. does the 'parking space' between expected entry and actual entry candles increase after a potentially filtered period (e.g. weekend).
Does 'price cross SMA' mean current 1H candle opened below SMA and passes or closes above SMA or
does it mean previous 1H candle close was below SMA and now the candle has passed the SMA or closed above the SMA.

Support has suggested introducing logging in order to determine what your code is actually doing - the line you've provided above doesn't exactly
scream 'judicious use of logging' does it? At the very least you can try to output the price, the time/bar, where your code is in its execution flow,
values when a trade is made, etc? o.O If you can output the price and sma you use in your comparison at least you have something you can
use to trace both the price and sma on the chart.

Support and the rest of the community are here for you but you have to help us help you...

Thanks CriticalSection for your answer . Yes I am trying to get some help because I have done VBA previously but Java it is completely new to me and I am getting familiar with it and OOP but I still have some issues...(I have done finance at the university).

But let's proceed with order:

a) I have setted filter to: Filter.ALL_FLATS;
b) smaTimePeriod = 200;
numberOfCandlesBefore = 200;
sma = indicators.sma(Instrument.EURUSD, Period.ONE_HOUR, OfferSide.BID, IIndicators.AppliedPrice.CLOSE, smaTimePeriod,indicatorFilter,numberOfCandlesBefore ,prevBar.getTime(),0)[0];
c) For the testing it is setted to 1H - Bid - 4 Ticks at OHLC;

I have noticed that changing the numberOfCandlesBefore variable yields some results. Infact putting it to 250 I get the first position opened properly but not the others (as you may see form the attached printscreen). I can't picture why since the smaTimePeriodi is the same: 200.

Yes I have to admit I don't really know what "logging in" means and what I have to do in order to fulfill Support's request.
I have tried to figure it out by myself but...bad times how can I do it?

Should I post my code?If there is anything else that I can do just tell me

Thank you for your help!

A quick check reveals one issue. You and I both assume that the array coming out of that SMA is in reverse order.
i.e sma[0] holds the most recent and sma[199] the most distant.
But it appears that the array you're getting out is actually in the traditional left to right ascending order.

sma[0] is the most distant candle you requested.
sma[199] is the most recent.

So that [0] you have sitting on the end there should be pulling the oldest value not the most recent.

If you use (we don't know) that to determine the cross to the right of the graph that's the culprit.

I implement all my indicators at the strategy level so I've not played around with Jforex functions to for
example pull the 10 period filtered SMA between time t0 and time t1 so this is new to me.

The prevBar.getTime() might be a different issue because it doesn't give you the SMA on current forming
candle, only up to the previous one (if indeed prevBar represents the last bar coming in to onBar).
The appropriateness of this depends on your actual strategy anyway.

Logging just means where you have code written that has a problem and really can't tell by looking where
it's going wrong, using Printout or Logging functions to output various variables (like the contents of that sma array)
or useful messages like "P was true so entering if", "Pi * D was greater than 60", etc.

This small example here told me about the organisation issue.

double [] test = null;
try{ test = i.sma(Instrument.EURUSD, Period.FIVE_MINS, OfferSide.BID, IIndicators.AppliedPrice.CLOSE,40, Filter.WEEKENDS, 5, teh.tick.getTime(),0); }catch(Exception e){}

// jforex arrays should match mine if they are also reversed 
System.out.println("SMA40[0]: " + test[0]); 
System.out.println("SMA40[1]: " + test[1]);
System.out.println("SMA40[2]: " + test[2]);
System.out.println("SMA40[3]: " + test[3]);
System.out.println("SMA40[4]: " + test[4]);

// my arrays are all reverse (0 is most recent) ordered 
System.out.println("mySMA40[0]: " + mysma[0]); 
System.out.println("mySMA40[1]: " + mysma[1]);
System.out.println("mySMA40[2]: " + mysma[2]);
System.out.println("mySMA40[3]: " + mysma[3]);
System.out.println("mySMA40[4]: " + mysma[4]);

output was

SMA40[0]: 1.29587975
SMA40[1]: 1.2958990000000001
SMA40[2]: 1.2959155000000002
SMA40[3]: 1.2959377500000002
SMA40[4]: 1.2959590000000003

mySMA40[0]: 1.2959590000000003
mySMA40[1]: 1.2959377500000002
mySMA40[2]: 1.2959155000000002
mySMA40[3]: 1.2958990000000001
mySMA40[4]: 1.29587975

Yes your hypotesis does make sense. I am going to check that out.
Excuse me for the dummy question but where should I get the output from?
I mean in Eclipse/NetBeans environment the output will appear in the Console tab but since
I am editing the code directly in the Dukascopy Workspace I can't find it anywhere...

Nobody knows?

From java console, see: viewtopic.php?f=81&t=39073