# Why do these Monthly vs. daily plots differ?

I got my data from Thomson Reuters Datastream. As an Input for my plot i calculated daily Returns based on the Return-Index provided by datastream. Then i plotted the Monthly and the daily Returns.(Same color = same Portfolio, Blue = Green - Red) Can differences like these occur because of data outliers, that are visible in the daily Return-Index and not in the monthly?

Also: Are there any screening recommendations regarding Datastream equity Data ? Does it make sense to exclude monthly Returns >300% if you evaluate emerging markets ?

EDIT

for daily Returns : (RI(t)/ RI(t-1)) -1 monthly Returns: (RI(endofmonth t)/RI(endofmonth t-1)) -1

• It's difficult to say without looking at the data. Assuming you've lined everything up properly and aggregated from daily to monthly correctly, the two should match. The two thoughts I have as to why they don't are potential issues in the underlying data, which seems unlikely to cause issues with several different securities. The second, depending on how daily returns are calculated and represented, is the possibility for gaps up or down between close and open that wouldn't be accounted for in the daily returns but would in the monthly return, and would cause discrepancies as a result. Commented Jul 18, 2019 at 18:54
• Screening (ie, stock selection) is a completely separate issue and you'd be best served to simply ask a second question to get input once you've cleared up your data issues. Commented Jul 18, 2019 at 18:55
• By looking at the chart, there are some obvious spots where the lines seem to 'jump' apart. I would look at your data at those points starting with the earliest ones first. Commented Jul 18, 2019 at 19:17
• This answer is quite late, however, it makes all the sense that your cumulative plot of daily data is not the same as a cumulative plot of monthly data. x * 1.05 is smaller than x* (1 + 0.05/20)^20. Regarding in or exclusion of monthly returns, I would cut the returns off for instance at a 50% level. Several papers before have done methodologies like this (e.g. two centuries of momentum)
– Bart
Commented Aug 2, 2019 at 11:58