BackTrader 1 : Concepts

Platform Concepts

Data Feeds - Passing them around

Strategy 안에 Data 를 어떻게 집어넣어야 할지 고민할 이유 없음

Lines

모든 object 는 lines object 로 통일되어진다

• Lines declaration
  • indicator 개발시 lines 추가로 간단히 indicator 만들기 가능
    •  
    • class SimpleMovingAverage(Indicator): lines = ('sma',) ...
• Indexing: 0 and -1
  
  • lines 라 불리우는 것들은 line series 이기 때문에 현재를 보고 싶다면 0 으로 해서 볼수 있다.

    • def next(self):
          if self.movav.lines.sma[0] > self.data.lines.close[0]:
              print('Simple Moving Average is greater than the closing price')

  • 전에 set points 로 indicating 을 한다면 -1 로 과거를 볼수 있다
• Lines Coupling
  • 다른 데이터로 조합된 indicator 를 비교할때 line coupling 을 통해 해결할수 있다.
  • Ex)
    • A daily data feed has around 250 bars per year
    • A weekly data feed has 52 bars per year

    • class MyStrategy(bt.Strategy):
          params = dict(period=20)
      
          def __init__(self):
      
              # data0 is a daily data
              sma0 = btind.SMA(self.data0, period=15)  # 15 days sma
              # data1 is a weekly data
              sma1 = btind.SMA(self.data1, period=5)  # 5 weeks sma
      
              self.buysig = sma0 > sma1()
      
          def next(self):
              if self.buysig[0]:
                  print('daily sma is greater than weekly sma1')

Operating the platform

Line Iterators

• The key to iteration, just like with regular Python iterators
  • The next method
    • It will be called for each iteration
    • Called when the minimum period for the line iterator has been met
  • But because they are not regular iterators, two additional methods exist:
    • prenext - Called before the minimum period for the line iterator` has been met.
    • nextstart - Called exactly ONCE when the minimum period for the line iterator` has been met.

Extra methods for Indicators

• 스피드를 빠르게 하기 위해서 runonce 기능을 사용할수 있다.
• runonce 는 data 에 직접 접근해서 한번에 한번에 처리한다
• The defined methods
  • once(self, start, end)
  • preonce(self, start, end)
  • oncestart(self, start, end)

Minimum Period

• sma = btind.SimpleMovingAverage(self.data, period=25)

• “period=25” instantiated moving average would have its methods called as follows:
  • prenext  - 24 times
  • nextstart - 1 time (in turn calling next)
  • next -  n additional times until the data feed has been exhausted

• sma1 = btind.SimpleMovingAverage(self.data, period=25)
  
  sma2 = btind.SimpleMovingAverage(sma1, period=20)

• What now goes on:
  • The same as above for sma1
  • sma2 is receiving a data feed which has a minimum period of 25 which is our sma1 and therefore
  • The sma2 methods are called as indicated:
    • prenext the first 25 + 18 times for a total of 43 times
    • 25 times to let sma1 produce its 1st sensible value
    • 18 times to accumulate extra sma1 values
    • For a total of 19 values (1 after 25 calls and then 18 more)
    • nextstart then 1 time (in turn calling next)
    • next the n additional times until the data feed has been exhausted
  • The platform is calling next when the system has already processed 44 bars.

Data Feeds

The platform provides several data feeds:

• Several CSV Format and a Generic CSV reader
• Yahoo online fetcher
• Support for receiving Pandas DataFrames and blaze objects
• Live Data Feeds with Interacive Brokers, Visual Chart and Oanda

A Strategy (derived) class

There are 2 methods which at least need customization:

• __init__
• next

A Cerebro

a Cerebro instance is what brings everything together and execute the actions

 

Defaults are taking care of if nothing special is wished.

• A default broker is created
• No commission for the operations
• Data Feeds will be preloaded
• The default execution mode will be runonce (batch operation) which is the fasterCustom indicators do not need to implement the runonce functionality. Cerebro will simulate it, which means those non-runonce compatible indicators will run slower. But still most of the system will run in batch mode.
• All indicators must support the runonce mode for full speed. The ones included in the platform do.