Test Memo 20161107
Goal:
- Test the intraday data.
Results:
- "Pipelines are meant to be computed on the order of once a day for security selection (the values in pipeline only uses daily data as recent as the previous trading day)" by Matthew Lee
- Reference: https://www.quantopian.com/posts/custom-factor-including-intraday-data - Attempt on intraday data running every minute or every n minutes:
- Reference: https://www.quantopian.com/posts/i-want-this-algorithm-to-only-run-handle-data-every-5-minutes-or-at-least-only-make-trades-every-five-minutes-how-can-i-do-this
Test Memo 20161106
Issues:
- Learn how to use Quantopian, especially pipeline feature.
- Migrate code of FIN315 final to Quantopian and cross-compare the results.
- Figure out the glitches between pipeline and data.history() when getting data.
Results:
- Utilize the official documents and the forum to learn:
- Quantopian Lectures: https://www.quantopian.com/learn
- How to Use Features from Quantopian: https://www.quantopian.com/tutorials
- Practical Concepts and Stategies (The content keep updating): https://www.quantopian.com/lectures
- Quantopian Forum: https://www.quantopian.com/posts - Pipeline set_screen by sid: https://www.quantopian.com/posts/pipeline-set-screen-by-sid
- Data sources:
- Quantopian’s internal stock price data basically keep consistent with Yahoo data
- https://www.quantopian.com/help
- https://www.quantopian.com/posts/pipeline-values-dont-match-history
- Note: Only for stock. Haven't checked the fundamental data yet. - Glitches of getting data between pipeline and datahistory():
- In pipeline, it executes everyday before the market open which mean it return the close price of last day. In data.history() method, it always returns current price. For example, data.history(,,5,'1d') returns [last_4d, last_3d, last_2d, last_1d, current_price]. Pipeline methods returns [last_5d, last_4d, last_3d, last_2d, last_1d].
- References: https://www.quantopian.com/posts/pipeline-values-dont-match-history
