Earnings comparison: relative deviation including negative values

Hi all, I am not sure if this is the right place to ask, but I hope one of you guys understands my challenge and can help me.

I want to compare past earnings with earnings projections for a number of companies. To be precise, the idea is to measure how "aggressive" or "conservative" earnings projections were in certain time periods, compared to the respective historical earnings.

For instance, if historical earnings (let's say, average of past five years) were 100$, and earnings projection is 150$, this is 150% of historical earnings. If earnings projection is 200$, this is more aggressive (or optimistic, if you like).

So far, so good. My problems start when negative values (=losses) appear, either in the past or in the projection. Let's make some examples:

hist. earnings ____ projection ________ perception
Company A _________ -100 ____________ -50 _________ aggressive
Company B _________ -100 ____________ 200 _________ extremely aggressive
Company C _________ -100 ___________ -200 _________ conservative
Company D __________ 100 ___________ -300 _________ extremely conservative

I have tried to find a measure to treat these values correctly but failed constantly. I guess I need a relative measure since my objective is to weight every data point equally. Does someone have an idea how to deal with this?

Thank you very much for your ideas in advance!
Last edited: