Hi Chris,

By default, the data in Fetcher is forward-filled. So your algo will buy 100 shares of AAPL every day from 1/6/2009-1/19/2009. You can easily fix this behavior in your CSV setup. Something like this:

date, symbol, shares, action
1/6/2009, AAPL,100, buy
1/7/2009, AAPL,0, no action
1/8/2009, AAPL,0, no action
1/9/2009, AAPL,0, no action
1/10/2009, AAPL,0, no action
... 1/20/2009, AAPL, 100, sell

You can then code your strategy to buy when the signal says BUY, sell when it says SELL and otherwise take no action. Here is a sample skeleton:

def handle_data(context, data):  
  for stock in data:  
      if data[stock]['action'] == 'no action':  
         log.info("Not trading today")  
          return

      if data[stock]['action'] == 'SELL':  
          log.info("Selling today")  
          order(stock, shares)

      if data[stock]['action'] == 'BUY':  
          log.info("Buying today")  
          order(stock, shares)  

Thanks Alisa.

I think your suggestion will work for the time period between my buy and sell, but I'm still left with an issue that the sell signal is filled forward after 1/20/2009 and I'm ending up with a growing short position after my sell. I think, however that your on the right track and that I'll need to modify my portfolio constructor to output a daily set of portfolio positions and target weights.

BTW, I made a temporary update to the code to work around this by using an order_target(stock, shares) call for the buys and order_target_percent(stock,0) for the sells. This way I only buy up to the target amount and always sell down to zero, however this will not work in the long run as my portfolio constructor does generate partial buy / sell orders.

thanks,
Chris

to keep from selling into short territory, why not always make the last line of your CSV indicate 'no action'?