A little advice....please.
I have managed to cobble together a python script that takes the Quandl WIKI export (1.7Gb) and reduces it down to 10 long buys and 10 short sales for each month. Enter the positions at beginning of the month, exit at the end...repeat. So, the .csv would look like (I'll include a few random stocks to illustrate):
date ticker ls 10/1/2000 AAPL long 10/1/2000 JNJ long 10/1/2000 MSF long 10/1/2000 NRG long 10/1/2000 A long 10/1/2000 OMI short 10/1/2000 CELG short 10/1/2000 ABBV short 10/1/2000 KMX short 10/1/2000 CSCO short 11/1/2000 SYMC long 11/1/2000 QCOM long 11/1/2000 PFG long 11/1/2000 LUK long 11/1/2000 EXPE long 11/1/2000 DLTR short 11/1/2000 STZ short 11/1/2000 COST short 11/1/2000 DVN short 11/1/2000 CSCO short ...etc, etc....
I am looking for some guidance on how to feed these in to backtest or, if it exists, another method of testing this as an equal weight portfolio. Since the table is a result of an algorithm, I need a way to backtest it as a dynamic, monthly portfolio. It is a challenge since the triggers are in the screening process. I would prefer to build the strategy in backtest but do not understand it well enough yet. Maybe contract someone to do it so the framework would be there then learn from that?
I think I was attempting to turn backtest into a giant stock screener and that is part of the problem.
Thanks for any input. So close yet so far away!
Probably related to the answer to this post a couple of days ago: https://community.backtrader.com/topic/678/manual-order-execution-trigger/3.
There is a post on how to automatically trigger orders at a given price for historical performance evaluation.