OPENEND in Test mode: Inconsistent effect on main chart

When plotting a Test result with OPENEND set:
Trades which are not exited by the end of the simulation are not plotted. (Expected).
The equity curve in the main chart still reflects them (Unexpected).

The net effect is that the Equity curve rises and falls mysteriously. There are no trades plotted on the graph associated with the changes in the plotted equity curve.

I am not sure how you should solve it. In my strategies, I will set OPENEND only in Train mode. Perhaps that should be default / recommended.

Hi GPEngine. I saw the same thing and reported it a couple of weeks ago - no reply as of yet.

Glad to know this time it wasn't just me!

Re. OPENEND in Train mode, I've been thinking that a strategy should via the Test process become such that OPENEND isn't significant; and then OPENEND in Training shouldn't matter. Does this make sense to you?

Thanks.

That is a good goal. But while you are developing strategies that only approach that goal, OPENEND is good for subtracting out the effects you eventually want to eliminate.

OPENEND works maybe not as expected, but at least as intended. The results of end trades do not enter the performance statistics, but as long as they are open, they do affect the equity curve, and when they cause a drawdown, it will also affect the performance.

Thanks jcl.

Perhaps the manual could be updated? "ignore" is a pretty strong statement and leads to the not included from any perspective understanding...

Also, could you explain a bit more about the intent of partially in / partially out? Is it meant to be pessimistic? Or was it just a lot more programming effort to go back and remove these trades from the equity curve as well?

Thanks.

Yes, I normally update the manual every time when such issues come up.

The purpose of ignoring the results of open-end trades, but keeping their effect on the equity curve is testing systems with long-term trades such as grid traders. Closing open trades at the end has a strong "end effect" on the performance, resulting in wrong performance statistics when the test period is not sufficient long.

However the effect on the equity curve must not be ignored, because it would also happen in live trading even when those trades are not premeturely closed.

Thanks jcl.