@vladisld Thankyou @vladisld it worked.

Hi,

I am trying to back test a 2 (or more) asset strategy which require selling one fully and buying other asset. I am doing this with cheat on open using next_open.

 def next_open(self):
        DEBUG_EN = 0;
        # Exit positions if any
        for i, d in enumerate(self.datas):
            dt, dn = self.datetime.date(), d._name
            pos = self.getposition(d).size
            if re.search(self.trade,dn):
                pass
            else:
                if pos:
                    self.close(data=d)
                    print(f"DEBUG :: NEXT OPEN :: {dt} Closing Position of {dn}")
                else:
                    pass       
        # Buy positions if any
        for i, d in enumerate(self.datas):
            dt, dn = self.datetime.date(), d._name
            pos = self.getposition(d).size
            if re.search(self.trade,dn):
                if pos:
                    # Position already exists so no need to do anything
                    pass
                else:
                    self.buy(data=d)
                    #cash = round(self.broker.getcash(),2)
                    #no_of_shares = self.getsizing()
                    print(f"DEBUG :: NEXT OPEN :: {dt} Opening Position of {dn}")
            else:
                # Do nothing as trade signal is to buy {dn}
                pass
        
                

What I am observing is that sizer is calculating the position as per leftover cash in folio without considering the cash outcome of sell transaction. I am using slight modified version of percent sizer worth 100% cash, modification to change open instead of close in next open. How can I make sizer take care of the cash generated from previous sell. Following are the debug messages I am getting from above code:

DEBUG :: NEXT OPEN :: 2008-01-25 Closing Position of asset1
DEBUG GETSIZING ::: Triggered asset2
DEBUG :: NEXT OPEN :: 2008-01-25 Opening Position of asset2
get_order_type :: Sell Order is Submitted
get_order_type :: Buy Order is Submitted
get_order_type :: Sell Order is Accepted
get_order_type :: Buy Order is Accepted
get_order_type :: Sell Order is Completed
get_order_type :: Cash Status : 102037.1
Notify Msg : Sell executed at price 53.9 on 2008-02-01. Total Value 99950.4
get_order_type :: Buy Order is Completed
get_order_type :: Cash Status : 102037.1
Notify Msg : Buy executed at price 11.3 on 2008-02-01. Total Value 45.2

Thanks and Regards
Ashutosh Ojha

@vladisld Thankyou @vladisld it worked.

@run-out Thanks for responding. I am still getting the same error. BTW my all other analyzers are working with _name itself.

Hi,

I am trying to get rolling returns for 1/3/5 year time frame on my weekly strategy. I am trying to use logrollingreturn analyzer for this purpose and trying to pass timeframe parameter as 52/52w etc, but getting error for the same. Can some one please help me providing details of how to use these parameter. Here is a code snippet.

cerebro.addanalyzer(bt.analyzers.LogReturnsRolling, _name='myrollingreturns' , timeframe=52)

Here is the error:
UnboundLocalError: local variable 'ph' referenced before assignment