3. Accessing data after finished run
For code/output blocks: Use ``` (aka backtick or grave accent) in a single line before and after the block. See: http://commonmark.org/help/

Accessing data after finished run

  • A

    Inside my strategy, I'm doing some algabraic caluclations and I want to be able to access those as arrays after my backtest run is finished.

    I tried creating an SMA(var ,period=1) but I still don't know how to access the data.

    I assuming it's somewhere in strategy.* but where? Any help?

    0

  • You run your strategy as

    result = cerebro.run()

    Then result object contains everything.

    Reference Returning the results section in Cerebro

    0
  • A

    I've read the page though it's not clear to me to get access to the data. The documentation says do it like so.

    thestrats = cerebro.run(tradehistory=True)
thestrat = thestrats[0]

    And then I've figured out how to get price data, trades, and orders

    # raw data
open = thestrat.data_open.array
high = thestrat.data_high.array
...

# trades
trades = [str(trade).splitlines() for trade in list(thestrat._trades.values())[0][0]]

# orders
orders = [str(order).splitlines() for order in thestrat._orders]

    But where I'm stuck now is how can I do this same thing with indicators?

    1
  • S

    @algoguy235 did have any luck figuring this out?

    0
    A
     1 Reply
  • B
    administrators

    Whereas orders are created by the system, indicators are created by end users. Something that could be my own code

    class MyStrategy(bt.Strategy):
    def __init__(self):
        self.my_smas = [bt.ind.SMA(period=x) for x in range(5, 30, 5)]

    Accessing the indicators later is a matter of getting thestrat as shown above and accessing the my_smas attribute (a list containing several Simple Moving Averages)

    1
  • A

    @backtrader. Yes thank you, dunno how I missed that!
    @søren-pallesen I did have some luck doing something in a sort of hacky way, but works exactly for what I need.

    class MyStrategy(bt.Strategy):
    def __init__(self):
        self.__special_array_for_later = [[None, 0, 0, 0,...]] #note nested brackets

    def next(self): 
        if some_condition:
            do_some_special_calculations()

    def do_some_special_calculations(self)
        self.__special_array_for_later.append([current_dt, calc1, calc2, calc3, ...])

    Then when it's all over I do this little move:

    special_array = thestrat._MyStrategy__special_array_for_later

    it turns out that thestrat._MyStrategy* has a bunch of tricks for debugging and other internals.

    From there I make my dataframe and do all the things I need.
    It seems to be working perfectly for the backtest environment. Not sure what's gonna happen when this all switches over to live 😬 but working good so far 👍🏼

    Another thing to note. I do my calculate AND append work in a separate function that gets called when some_condition is true. I do that because I don't want to weigh down my next(self) function with unnecessary code. Then when the run is over, I line up the special_array with my other rawdata and trades arrays using dt selections in pandas - during the data pre-processing and analysis steps. I don't know the exact overhead I'm saving by doing this. If I were to use indicators, they would calculate every cycle, but would be easier to line up after. I opted to save the overhead, and do the alignment later. But I imagine both would work.

    2