Note log file (and the dates it reports are the dates I decide to trade on but for all dates in the backtest XLI does not have the property):

2002-04-04handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2002-07-02handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2002-10-01handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2002-12-31handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2003-04-02handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2003-07-02handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2003-10-01handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2003-12-31handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2004-04-01handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2004-07-02handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2004-10-01handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2004-12-31handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])
2005-04-04handle_data:115DEBUGWARNING: rank not found for ticker Security(19657 [XLI])

There's some forward-fill stuff going on here, but I haven't had time to dig in. I will do so and get back to you.

Ahh -- then its probably related to some other issues I've been having (https://quantopian.com/posts/help-with-fetcher-and-filling-nan-values - sorry for the double post). Seems like no matter the fill method its always forward filled :-/ although in this case there is no data for XLI so that is bizarre.

Anyway, thanks a bunch for helping me out with this Dan. It is immensely appreciated.

I'm going to politely bump this thread. This problem specifically is hard to resolve. I can't fully understand why data are being filled at all in this data frame because there shouldn't be any NAN entries.

In this fetcher problem - https://quantopian.com/posts/help-with-fetcher-and-filling-nan-values - I've got a few ideas for some work-arounds but my feeling is that the issues are related. (Note I use fillna(0) but 'ffill' is used instead, maybe this is a default of post_func and is not getting changed?)

I continue to appreciate the help of all the staff at Quantopian.

Sorry Daniel. Haven't figured this out, but haven't forgotten either. Enlisting help.

Hi Daniel - I started looking into this problem by creating a minimal repro with just a record statement. Surprisingly, the code worked without any issue and plotted as expected, without any log lines. Here's the code for reference:

import datetime  
import math  
import numpy  
import pandas

def clean_col(df):  
    # df = df.fillna(0)  
    # df = df[['rank', 'sid']]  
    log.info(' \n %s' % df[df['sid'] == 19657].head())  
    return df

def initialize(context):

    #  
    #  
    #Read universe  
    context.SPY=sid(8554)  
    context.Sectors=[sid(19662), sid(19659), sid(19656), sid(19661), sid(19655), sid(19658), sid(19660), sid(19654), sid(19657)]  
    # context.Sectors=[sid(19657)]  
    #Read the rankings  
    fetch_csv('https://dl.dropboxusercontent.com/s/7shk9xm1xi311at/Ranks_QuantopianTest.csv',  
              date_column='Date',  
              post_func=clean_col)

def handle_data(context, data):  
    if 'rank' in data[sid(19657)]:  
        record(signal=data[sid(19657)]['rank'])  
    else:  
        log.warn("No rank for XLI")  

Looking at the log line in the code, I realized this is a whitespace problem. The else was outdented by one level, and so was being called at the conclusion of the for loop. Since XLI was the final element in your context.Sectors list, it was always being logged. Attached is a patched version with the desired indentation, and I was able to run a backtest without any of the log warnings or errors.

thanks,
fawce

@John - Duh, well I feel dumb for making that mistake. Thank you, sir. I was certain that the problems I was having here were related to the other issues I was having with filling nan values - https://quantopian.com/posts/help-with-fetcher-and-filling-nan-values - but I guess not. Do you think perhaps you can look at the other issues? I'd be much obliged.