Donchian Channel Breakout Strategy

Really cool!  This would seem like doing the same thing as something like this correct?
 

# region imports
from AlgorithmImports import *
# endregion

class DonchianChannelBreakoutAlgorithm(QCAlgorithm):

    def Initialize(self):
        self.SetStartDate(2020, 1, 1)
        self.SetEndDate(2022, 8, 1)
        self.SetCash(100000)
        self.symbol = self.AddEquity("XOM", Resolution.Daily).Symbol
        self.EnableAutomaticIndicatorWarmUp = True
		self.high = self.MAX(self.symbol, 20, Resolution.Daily, Field.High)
		self.low = self.MIN(self.symbol, 20, Resolution.Daily, Field.Low)
		
	def OnData(self, data: Slice):
        if self.symbol not in data.Bars:
            return
        bar = data.Bars[self.symbol]

        if bar.Close > self.high and not self.Portfolio[self.symbol].IsLong:
            self.SetHoldings(self.symbol, 1)
        elif bar.Close < self.low:
            if self.can_short and not self.Portfolio[self.symbol].IsShort:
                self.SetHoldings(self.symbol, -1)
            if not self.can_short and self.Portfolio[self.symbol].IsLong:
                self.SetHoldings(self.symbol, 0)

Hi Axist,

Very close! If we wanted to use Min/Max indicators, we would need to add a delay period via indicator extensions.

self.high = IndicatorExtensions.Of(Delay(1), self.MAX(self.symbol, 20, Resolution.Daily, Field.High))
self.low = IndicatorExtensions.Of(Delay(1), self.MIN(self.symbol, 20, Resolution.Daily, Field.Low))

Otherwise, 

bar.Close > self.high.Current.Value

and

bar.Close < self.low.Current.Value

would always be false. The Donchian Channel indicator has this delay built-in.

See the attached backtest for reference.

Best,
Derek Melchin

Want to invest in QuantConnect as we build the Linux of quant finance? Check out our Wefunder campaign to join the revolution.

Hi, 

I'm a new user and still finding my feet. I've tried cloning Derek's DonchianChannelBreakout script but can't seem to getting working in the backtest. The script runs without errors but the backtest returns no trades and a flat equity line. I've tried swapping out the original “XOM” ticker for “SPY” and “QQQ” but get the same results. I'm sure the answer is something simple that I can't yet see and would be grateful if more experienced eyes could point it out. 

Thanks! 

Hello PaleMountainBird  

There is a small logic error in the implementation of the code.

DCH indicator runs with a simple logic design:
UC = Highest High in Last N Periods
Middle Channel=((UC+LC)/2)
LC = Lowest Low in Last N periods

When you check the condition:

if bar.Close > self.indicator.UpperBand.Current.Value

here self.indicator.UpperBand.Current.Value is updated with the latest value and when we check the condition it returns a boolean false everytime resulting in no trades.

A simple way to correct the logic is to store the value of "self.indicator" in a variable as previous value and then check this with the current close price. This will give you a correct backtest that will match the original backtest attached by Derek.

Check the backtest out and let me know your reviews!

Hello PaleMountainBird 

There is a small logic error in the implementation of the code.

DCH indicator runs with a simple logic design:
UC = Highest High in Last N Periods
Middle Channel=((UC+LC)/2)
LC = Lowest Low in Last N periods

When you check the condition:

if bar.Close > self.indicator.UpperBand.Current.Value

here self.indicator.UpperBand.Current.Value is updated with the latest value and when we check the condition it returns a boolean false everytime resulting in no trades.

A simple way to correct the logic is to store the value of "self.indicator" in a variable as previous value and then check this with the current close price. This will give you a correct backtest that will match the original backtest attached by Derek.

Check the backtest out and let me know your reviews!