Dear Dukascopy, this is fatal error and it makes historical tester results wrong on Renko bars.

Example:
EUR/USD 10 pips BID Renko bar
starts at 2013-12-31 10:56:51:314
ends at 2013-12-31 13:03:02:500
OHLC is 1.377 / 1.377 / 1.376 / 1.377

onBar from Renko feed is received in strategy at 13:03:02
but this bar was formed by price 1.378 at 13:13:37
so I have informations from 13:13 at 13:03...

I debuged your code at "RenkoCreator.class" at function "analyse" is input object TickData (tick at 13:13)
This formed the Renko bar and onBar is posted to strategy at 13:02 what makes see future...

Dear Dukascopy, are you working on this bug?

The issue has been registered and will be fixed as soon as available.

After preliminary investigation we did not manage to replicate this. Could you please provide a concrete case - an example strategy and a printscreen of you Historical Tester settings?

You need only subscribe to renko feed and compare onBar with onTick.
Next you must start longer time range probably, because is it affected by ticks/bars caching, sometimes it not appears on short backtest. Try Last Year backtest.

It looks like it appears often when you moving chart (probably it slow down backtest and do more caching). For example, when you do several move chart and then press automatic chart shift, chart is automatially adjusted to correct bar, but there are future bars or right side out of visible chart screen, if you scroll there.

Maybe it is depends on "Visual mode" checked? I tested with visual mode.

I'm sure this is bug, because backtest results are unreal profitable...

Please provide:

• The example strategy that you use.
• Printscreen of your HT settings.
• Time point when the bug occurs and what you manually do with the chart to invoke the false behaviour.

Times are in first message. It can occur in different times, depens probably how fast or slow is computer/caching. I not remember exactly when long time range backtest was with or witout chart, but if yes, in this case I do nothing with chart. As I said, problem is in your cache mechanism, bars are built in advance and sent to the onBar early.

The strategy that you printed does not print anything. We reiterate that in order to track down the possible bug we would need a time point where the error takes place. Thus the proper case would require a screenshot of:

• Your strategy messages that simply print the feed (you can use an adjusted version of https://www.dukascopy.com/wiki/#Chart_fe ... _to_a_feed).
• Historical Tester settings.
• The chart with the OHLC widget with the mouse pointing to the feed element that you find being wrong.

The image could be similar to this one (except that you would follow the feed information, not the indicators):
https://www.dukascopy.com/wiki/#Add_indicators_on_chart/Include_in_OHLC

Dear Dukascopy, I debugged your code and wrote you where is problem. You can give this information directly to programmers and they can find problem, I'm programmer and with this information I'm able locate problem. Problem can be very easy described:

The onBar for Renko in Historical Tester is received in advance, earlier than it should. That's all what your programmers need know to find bug. Only "preliminary investigation" is not enough.

Backtest of simple strategy returns 99% profit/profitable trades and looks like holy grail, as you know sure, this is not possible. I have my self analyze software and I know such us result is always bug in platform. More I debugged it is problem in your platform. I have developed the strategy to order, so I cannot give you the strategy, I can ask the customer.

I told you about your bug and I'm sorry, I not have more free time to do job as your programmer or tester and create strategy for testing with debug messages.

But if you need help, you can hire me.

To proceed we need to be able to reproduce the problem, which by the information that you had given we did not manage to do. See our tester settings and the strategy:


Thus, please provide additional details on the reproduction steps.

Dear Dukascopy, you are not able reproduce problem as user, because you not know about bar was built based on future data. You need go to JForex platform source code and debug it. I wrote in first post where your programmers can start find problem.

The first step is to reproduce the problem, which we were not able to do by the information that you have provided up till now. Please provide the problem reproduction steps.

Dear Dukascopy, I told you you are not able reproduce this kind of problem as user, there is nothing to see by eyes what it wrong, and I not have access to your source code to tell you more. If you are not able take it, please forward my information to IT department supervisor, thank you.

If you can't see the problem on charts and can't detect it from a strategy, then what is the case here and how to reproduce it? Your suggested debugging stops only one thread, while the others continue their work, so if you stop renko-creating thread then obviously the tick feed will go on and tick times will be greater than the last-formed renko brick end time.

When tick time will be greater than last renko, this is ok.
But problem is, actual tick time is smaller time than renko posted to strategy, so you see renko from future. (I told it several times)
I not study what times are used on Renko bars, so I can tell you exactly, this time is wrong
(and this is lot work to decode your renko bars process and tell you exactly, becouse renko bars are dealyed, last is not visible etc..)
Problem is not in creating bars, problem is, WHEN bars are posted to onBar function in strategy.
So you can debug onBar and onTick together with renko-creating thread,
and you will see tick from 13:13:37 create bar which is posted to onBar at 13:03:02
what is 10 minutes in advance, and this is what tell me when will be market after 10 minutes.
This is what is in first post and what is information needed to correct problem.

Debugging a multi-threaded program by definition will change its behaviour. Please provide a case and reproduction steps which do not involve debugging the platform.

Dear Dukascopy, you are wrong, you can debug it if you understand how engine works.
Please stop request me for working for you free, it is a lot of work to prepare reproduction steps.
I not have more time help you, please provide serious bug to more sophisticated department, thank you.
Error may have traders thinking they have a profitable system which will be live unprofitable.

In attachment is pdf report and equity from pips, from strategy with reaction to last renko bar
(up renko is buy/down renko is sell -> and here occurs the bug, because bar direction tells real market direction in future,
because is received in advance) with SL/TP as 10/10 pips and linear winning ratio about 80%...
Right equity for this simple strategy must be down direction no profitable equity and not up.

Dear Dukascopy, are you working on this bug?

We cannot reproduce the problem. Last tick time is never smaller than the Renko bar end time. Please run this strategy attached and show us the opposite.

Renko bar probably have right time, but was created from future ticks.
YOU CANNOT REPRODUCE IT AND CATCH IT IN STRATEGY, THERE IS NOTHING TO SEE, YOU MUST DEBUG IT!
I posted you exactly where you can debug it in first post.
Place break point A in your source code in function "analyse". Check that specified bar X is created from tick with specified time Y.
Place break point B in renko onBar function in strategy. You will see, onBar X is received in time before time Y.
That's all! You must debug it!