Price is null in certain occasions

[rpreosck]

I'm optimizing some parameters, and some configurations result in a null price in the middle of the optimization. I'm not sure what is causing it, maybe trying to sell twice in the same period somehow

Steps to Reproduce

I use this Function:

stats, heatmap = bc.optimize(
    iaf1 = [0.015],
    maxaf1 = [0.16],
    mgain = [0.02,0.03,0.04,0.05,0.06,0.07,0.08,0.09,0.10],
    mloss = [0.01,0.02,0.03,0.04,0.05,0.06,0.07,0.08,0.09,0.10],
    
    constraint=lambda p: p.iaf1 < p.maxaf1,
    maximize='Equity Final [$]',
    return_heatmap=True)

WIth a parabolic SAR indicator:

    def next(self):
        if not self.position:
            if self.boole == 1:
                self.buy()          
            elif self.boole == -1 and not self.position:
                self.sell()
        else:
            if self.position.pl_pct >= self.mgain or self.position.pl_pct <= -self.mloss:        
                self.position.close()

Getting this assertion error

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in sell(self, price, sl, tp)
    465 
    466     def sell(self, price=None, sl=None, tp=None):
--> 467         assert (tp or -np.inf) <= (price or self.last_close) <= (sl or np.inf), "For short orders should be: TP ({}) < BUY PRICE ({}) < SL ({})".format(tp, price or self.last_close, sl)  # noqa: E501
    468         self.orders._update(price, sl, tp, is_long=False)
    469 

AssertionError: For short orders should be: TP (None) < BUY PRICE (nan) < SL (None)

Where the price is null, but the model works fine on other parameters.

[kernc]

If you can compile this up into a minimal working example that reproduces the issue, I'd be happy to investigate it.

[kernc]

self.last_price, which is printed out as nan in your example error, is never written to. The null value for closing price must have come from the input data itself. 👀