wrong logic of buy and sel

[Realmlai]

Expected Behavior

I found the buy and sell signal was wrong.
please check out the attachment.

Actual Behavior

wrongly P/L -1.538
because the trade is a win trade

Steps to Reproduce

1.
2.
3.

Additional info

• Backtesting version:

[kernc]

The green up arrow is not a buy signal, but rather an end point marker of a market-long (upwards) and profitable (green) trade.

Note, Strategy.sell() implicitly closes the long position and opens a new short one (that'll change in the upcoming version, see #47). If that doesn't answer your concerns, why do you think the last trade is a win?

[Realmlai]

Great, thank you kernc

[MattoDemente]

Then, how do I identify the buy/sell signals in the plot ?

[kernc]

The dotted lines indicate duration of trades and the PL triangles closing trades' success. There are no other visual indicators showing trades entering, nor order placing. It's very simple to devise something yourself:

class MyStrategy(Strategy):
    def init(self):
        # Shows buy/sell signal as a scatter plot. Init to all-NaN.
        self.signal = self.I(lambda: np.repeat(np.nan, len(self.data.Close), scatter=True)

    def next(self):
        ...
        # on some condition
        self.buy()

    def buy(self, *args, **kwargs):
        self.signal[-1] = 1  # On buy, input 1 instead of existing NaN
        super().buy(*args, **kwargs)