Hello,
I am trying to put together my first "real" backtest, and would appreciate some help on figuring out how to design the system:
For each day, I have about 100 ASCII files containing minute-bar data for stocks deemed "interesting" by an external process (SEC filing data/etc). This selection changes for each day (so for 1 year, I have about 26,000 files). I can load these into the QuoteManager via ASCII import.
For each security, I should be able to use standard MAVG style indicators to determine if I wish to buy and sell each.
How to best backtest this? I am thinking I should use the Portfolio Trader system, but can I somehow dynamically modify my instrument listing based on the backtest date?
Thank you.
"Best Practice" to backtest a dynamic portfolio strategy?
- ABC
- Posts: 723
- Joined: 16 Dec 2006
- Location: www.abctradinggroup.com
- Has thanked: 125 times
- Been thanked: 410 times
- Contact:
Re: "Best Practice" to backtest a dynamic portfolio strategy
novaleaf,
what about simply excluding symbols from your code calculations depending on the date?
So for example you have a list of 1000 symbols in the Portfolio Trader and each day your code would read a list of symbols to use for that particular day. If a symbol is not in this list you would stop further code computations after the initial "if currentsymbol is contained within the list" check.
Regards,
ABC
what about simply excluding symbols from your code calculations depending on the date?
So for example you have a list of 1000 symbols in the Portfolio Trader and each day your code would read a list of symbols to use for that particular day. If a symbol is not in this list you would stop further code computations after the initial "if currentsymbol is contained within the list" check.
Regards,
ABC
Re: "Best Practice" to backtest a dynamic portfolio strategy
Hi ABC,
Thank you so much for the suggestion! I'm a bit worried that loading up 25k unused securities might have a performance hit... However I can't figure out a better approach in my stabs at the documentation, so I will try your approach.
I will write back informing if it turns out to be (in)feasible!
Thank you so much for the suggestion! I'm a bit worried that loading up 25k unused securities might have a performance hit... However I can't figure out a better approach in my stabs at the documentation, so I will try your approach.
I will write back informing if it turns out to be (in)feasible!