Goal: Compute average true range using ATR() in TA-Lib
Big O notation: Execution time and space trade-off when computing average true range
Solution 1:
- Call data.history() and data.current() in handle_data()
- Combine data
- Compute ATR()
Solution 1 saves space at the expense of execution time because data.history() returns the same results for each call to handle_data().
Solution 2:
- Call data.history() once before market open or using schedule_function()
- Call data.current() in handle_data()
- Combine data
- Compute ATR()
Solution 2 saves execution time at the expense of space because the results of data_history() need to be stored in memory.
Solution 2 is not feasible because context has limitations:
- Scalar value saved to context in initialize() is accessible in handle_data()
- Scalar value saved to context in before_trading_start() or schedule_function() is not accessible in handle_data()
- 3-dimensional array saved to context in before_trading_start() or schedule_function() is not accessible in handle_data()
Uncommenting line 121 throws KeyError: 'prices'.
