This repository features the data and methods used that leads to the graphs and conclusions of this Medium article
In this repository you are able to find the functions that are used to do the calculations as well as the pickles that are used to generate the graphs. As the full analysis took several hours, I created pickles to be able to properly store the results in an efficient way.
The table below explains the methods used and includes an explanation for any adjustments to the data.
| Variable | Description |
|---|---|
| Data | The data is gathered from Yahoo Finance via the Python package yfinance. Here the Adjusted Close prices are used for the simulation to prevent not taking into account potential dividends or stock splits. |
| Period | I have taken a 10 year period starting from 2010-09-27 until 2020-09-23 where I filled any missing values up to 3 days and dropped any tickers that still had NaNs afterwards. Furthermore, all date gaps like weekend days are included and forward-filled to prevent skipping months. Thus the assumption is made that you trade one day earlier if the selected date is a weekend day. I find this to be a reasonable assumption to take. |
| Tickers | The tickers used are all tickers from Yahoo Finance's World Indices page with the exception of 'IMOEX.ME', '^JKSE', '^TWII', '^IPSA', '^TA125.TA', '^CASE30', '^JN0U.JO' that were not included due to NaNs or not enough data available for the full 10 year period. |
| Trades | Trades occurred on the specific day and are assumed to happen straight away. This means you are also affected by the return that day. This results in a total of 120 trades in a 10 year period. I make the assumption that you are able to invest the full amount even though the ticker is priced higher. As there are many types of ETFs available for the indices with different (lower) prices it is reasonable to assume that you can consugmame much of the $100,- you invest. |
| Returns | Returns are calculated by the "New minus Old" principle. You invested a total of 120 * $100,- which is compared to the accumulated value at the end of the 10 year period. |
| Graphs | The graphs aggregate the returns and calculate the median values to give an average of the total return in the 10 year period. Quantiles are then calculated based on these median returns. |