3. Setting a commision changes buy size to be in monetary units?
For code/output blocks: Use ``` (aka backtick or grave accent) in a single line before and after the block. See: http://commonmark.org/help/

Setting a commision changes buy size to be in monetary units?

  • G

    Hi,

    When I use:
    cerebro.broker.setcommission(commission=0.005)

    and place a buy order:
    self.buy(data=self.data0, size=1)

    I see in 'notify_order' when the order status is order.completed:
    order.executed.price = 1.8768
    cerebro.broker.getvalue(self.datas) = 1.9732
    self.data0[0] = 1.9732

    which indicates that 1 unit of the stock was bought.
    However, when I set
    cerebro.broker.setcommission(commission=0.005, commtype=bt.comminfo.CommInfoBase.COMM_FIXED)
    and use the same buy order I get when the order is completed

    order.executed.price = 1.8768
    cerebro.broker.getvalue(self.datas) = 1.0
    self.data0[0] = 1.9732

    which indicates that the order was places to buy as many units as possible for 1 monetary unit.

    Is this a bug or a feature?
    How can I set a fixed commision in monetary units and still place buy orders where the size parameter indicates units of the stock to buy?

    Thanks!

    0
  • B
    administrators

    You are comparing apples and oranges actually.

    @Gil-Keren said in Setting a commision changes buy size to be in monetary units?:

    cerebro.broker.setcommission(commission=0.005)

    Here you don't specify the commission type and the original behavior is to consider you are operating on stocks with a % commission, because you haven't specified a margin (which is meant for future-like assets)

    @Gil-Keren said in Setting a commision changes buy size to be in monetary units?:

    cerebro.broker.setcommission(commission=0.005, commtype=bt.comminfo.CommInfoBase.COMM_FIXED)

    In this case you set commission type and the rest of parameters are taken at face value. The default is to operate with future-like assets.

    To understand the behavior read the parameter explanation for commission schemes

    0
  • G

    Thanks.
    I read the link you attached for the documentation.
    Note that the size parameter in buy() is interpreted as face value even I set a percentage commission:

    cerebro.broker.setcommission(commission=0.005, commtype=bt.comminfo.CommInfoBase.COMM_PERC)
    self.buy(data=self.data0, size=1)

    Then in notify_order:
    order.executed.price
    1.8768
    order.executed.comm
    0.009384 # This is 0.5% of 1.8768, makes sense so far
    cerebro.broker.getvalue(self.datas)
    1.0 # size parameter in buy() was interpreted as face value, not as amount of shares t buy.

    But my question is the following:
    How do I set the commission to be in in fixed monetary units, but still have the 'size' parameter in 'buy()' interpreted as the amount of shares to buy?

    0
  • B
    administrators

    The size value is always the number of shares, contracts or whatever you want to buy.

    Even if you checked the link, it is still being used wrongly. You are not buying shares, you are buying contracts. Either set stocklike=True or use the broker method addcommissioninfo and add an instance of for example: CommInfo_Stocks_Fixed

    Of fix the parameters of your own subclass like here: Docs - Extending Commissions

    0
  • B
    administrators

    In any case:

    • getvalue doesn't give the actual size. It gives you the value of your position which is not the size.

    To see the size, use for example:

    • self.position.size in the strategy (this is good for the 1st data in a multi-data approach and single data approaches)

    The value is given as 1.0, because in the absence of meaningful values to setcommission, the system assigns a value of 1.0 to the margin required to keep a future alive (as explained above, you are operating with future-like assets) and that's the value of your position (using the nominal value would in most cases exceed the total value of the account what would probably be puzzling)

    0
  • G

    Works like a charm, thanks for the detailed answer!

    0
Log in to reply
 
});