andrewkdinh/fund-indicators
1
0
Fork 0
mirror of https://github.com/andrewkdinh/fund-indicators.git synced 2024-11-22 04:44:18 -08:00
Code Issues Releases Wiki Activity

Create why-overhaul.txt

Browse Source
This commit is contained in:
Andrew Dinh 2019-01-30 23:54:27 +00:00 committed by GitHub
parent 5f63eeb57e
commit e644078142
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 29 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
why-overhaul.txt Normal file
View File

@ -0,0 +1,29 @@
1/30/2019
I decided to do an overhaul of this project for a few reasons.
1. I realized I don't need two separate classes (StockData & StockReturn).
I can just make one class (Stock) that holds all the values.
2. I realized how it was much easier to work with datetime than with strings.
In the overhaul, I'll be working with datetime as much as I can, only switching to strings when it's absolutely necessary.
3. I've found other API's/sources for stock data like pandas-datareader (https://pandas-datareader.readthedocs.io/)
I'll likely be sticking with the API's I have right now (Alpha Vantage, Tiingo, & IEX), but will keep these sources in mind in the case I need them.
In particular, I may be integrating the fix-yahoo-finance (https://pypi.org/project/fix-yahoo-finance/) module because Yahoo supports mutual funds.
4. My code is getting more and more cluttered the further I go into this project.
Therefore, it would be to my advantage to clean it up before I go onto the next step of analyzing the data.
I'll be focusing on having more functions (def) so it is more readable.
I would also like for my project to be easily adaptable into other projects should I have the need.
Fixing my code should help me do this.
5. As of now, I have not been calculating return and close values as I probably should have.
I'll have to do some more research into this, but this is what has transpired up to this moment:
As of now, I have been gathering data from multiple sources, averaging the close values for the same dates, and calculating return based on the final list.
However, I realize that adjusted close values can vary greatly, even with the same API.
For example, the same date from Alpha Vantage's monthly-adjusted close values is slightly different from their daily-adjusted close values.
Instead of this current process, I'll consult with a mentor to figure out if it is advantageous to:
1. Stick to one source
2. Use as many sources I can and only average the very final results
(Ex: use the results of unadjusted return from Alpha Vantage and results from Yahoo to average the two)
3. Choose a few sources and stick to just those.