I have N financial series consisting of 20 yearly revenue values each. In each series, there is a comparable trigger occuring during one year, but it doesn't necessarily occur in the same year across them. I want to know if we, on average, can identify revenue trends post this trigger. My current take is to "shift" each series so that they align on the trigger year. I then normalize each series by subtracting the mean and dividing by the STD. I average over all the series and plot the results.

Now, as each series is normalized, the result isn't interpretable in "raw" revenue figures, but is there some way to meaningfully calculate the average percentage change/deviation? I want to be able to say, for example, "post trigger, the average company experiences a 20% decrease in revenue". (I don't want to argue causation, but rather whether there is a pattern)

Should I perhaps convert the original series to one of percentage change and then sum/plot? Or is it meaningful to talk about percentage change on the average of N normalized series'?

My question is similar to: Is it okay to take the average of several normalized time series? but it didn't get any answers.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.