Lending Club Optimizer
Predicts the default rate of loans and automatically order the model's favorite currently available loans. Likely to increase ROI from 7.5% (Lending Club Average) to 13.5%. This project identifies investment stratagies that fit custom investment profiles. The user specifies the amount of capital she is willing to invest and either her desired expected return or her maximum risk tolerance.
Overview
- The
appdirectory has two major entry points.execute_orders.pywill order the model's favorite loans.process_apisupports this process.main.pyrenders a webpage that allows the user to view currently available loans and its expected rate of return.
scriptscontains code used in developing the modelnotebookscontains Jupyter Notebooks that tell a story of the process I followed
Replication
Getting the data
The full data is available to Lending Club members. The data can be downloaded here. It is critical to first log in as FICO scores are only available once logged in.
Access to the Lending Club API
You need to be registered on the Lending Club site in order to use the lending club API. I recommend that you store the API credentials as an environemntal variable for security reasons.
My app expects the name of the environmental variable to be LENDING_CLUB_API.
Starting the dashboard app
bokeh serve --show app
Notes on relative imports
It seems like it is going to be a better idea to handle things
Restructuring this app
The API here was somewhat ill-planned. This should be turned into a library that accesses multiple things and has a nice API that will perform various operations (orders, show current notes, and
Refresh TImes
2AM, 6AM, 10AM and 2 PM