Why Returns May Be Normalized in Time-Series Analysis
Summary
The document examines whether returns should be normalized after converting price observations into returns. The questioner notes that returns are already relative changes and asks what further normalization by a measure of prior performance accomplishes, particularly in time-series modeling. The answer does not explain a return-normalization procedure or connect it to autoregressive models.
Instead, the response discusses normalization of company financial statements: adjusting out unusual, nonrecurring expenses or gains to present a view of ordinary operating performance. It says the respondent personally does not normalize returns, while acknowledging that the course's approach may have a purpose. No data, definitions of the proposed denominator, or comparative results are provided, so the exchange does not establish when normalized returns are useful. Readers should distinguish the financial-statement practice described in the answer from statistical scaling of return series.
Key ideas
- The question distinguishes ordinary returns from a further normalization using prior performance.
- The answer discusses removing unusual items from company financial statements rather than normalizing return data.
- The response gives no method or evidence for applying normalization to time-series returns.
- The distinction between financial-statement adjustments and statistical scaling remains unresolved.
Tags
Full text
# Does it make any sense to normalize returns? # Does it make any sense to normalize returns? I have been going through a course for Time Series Analysis. First we learned to make returns from a time-series of stock index by (Xt - Xt-1)/Xt-1 . This makes the series stationary, which means we can run the Auto-Regressive model. It all makes sense until here. Than the returns are being normalized, dividing them by some previous performance of the specific index. Now this seems pretty incredulous to me. Also it is accompanied by the statement "They account for the absolute profitability of the investment in contrast to prices." Makes no sense to me. Returns are relative by themselves and than when they are further normalized they are relative to some previous value as well. My question is does it make sense to use relative returns and if so in what type of situations? ## Answer by shantanujoshii (score 1, accepted) https://quant.stackexchange.com/a/69273 As far as I know we use Normalization in reviewing the Financial Statements(PnL,Balance Sheets,Cash Flow) of a Company. Now What is that exactly? If the company is seeking external funding, normalized financial statements provide the investor or lender with a clear picture of the actual expenses, revenues, earnings and cash flow of the company during a particular period. When adjusting the financial statements, you should only remove outliers (discretionary expenses and one-time gains or expenses) that are unrepresentative of normal operating expenses of the business so that you get the clear picture of the performance of the company. I hope you get the intuition behind it and personally I also never normalise returns when I calculate it because it does not make much difference but what your Course is doing is something intriguing so it was worth answering this question. Thank you!
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.