The motivation for building this was the desire to view quick summary info regarding my Crypto-currency portfolio on Coinbase. I wanted to know how much money I put in and how much my portfolio was worth in addition to what accounts contributed how much. I was not able to find an easy way to do so with Coinbase directly, however fortunately they have a nice API.
One of the main goals (which is yet to be achieved) is to make it easy to see how each of the cryptos I bought is performing and how much it's contributing to the overall picture. I will also include general historical statistics which will hopefully let me better monitor what I may have done right and what not so much.
Of course, no assitant like this would be complete without a tool to set up automatic buys (or maybe that's a bit dangerous), but at the very least a system to notify me when a certain currency has passed pre-set thresholds. Too many times have I lost money because I missed an opportunity by a couple of hours.
To run this for yourself you will need to clone the repo by opening a starting up a shell and running
git clone https://github.com/djvaroli/coinbase_assistant.git coinbase_assistant/ && cd coinbase_assistant
You can then create a virtual python environment and install all the required dependencies
python -m venv venv/
source venv/bin/activate
pip install --upgrade pip
pip install -e custom/
pip install -r calypso/requirements.txt
Once that is done you will need to create a .env file where you can add your twilio and coinabse API keys. You also need to specify the version for docker-compose, but that can be set to anything.
It should look something like this
COINBASE_API_KEY=<YOUR_COINBASE_API_KEY>
COINBASE_API_SECRET=<YOUR_COINBASE_API_SECRET>
TWILIO_ACCOUNT_SID=<YOUR_TWILIO_ACCOUNT_SID>
TWILIO_AUTH_TOKEN=<YOUR_TWILIO_AUTH_TOKEN>
TWILIO_PHONE_NUMBER=<YOUR_TWILIO_PHONE_NUMBER>
DEFAULT_TO_PHONE_NUMBER=<YOUR_DEFAULT_TO_PHONE_NUMBER>
VERSION=1.0
The phone numbers are set just so that twilio has some default values for sending text messages.
If you have Docker installed, to startup the server simply run
# from inside the root directory
docker-compose up
This will make the backend component of the assistant available at http://localhost:80.
If you do not have Docker installed, or you want to run the service in debug mode simply run
uvicorn calypso.app.main:app --reload
If you do not want hot-reloading to be enabled simply drop the --reload flag.
Currently I have a set up that fetches the portfolio breakdown every 5 minutes. This job is set up using crontab. If you would like to enable that, there is a script that will handle it for you. Simply run
bash scripts/set_up_cron_job.sh
- Currently, I am planning to add a front-end interface with some added features, intuitive visualizations and historical statistics to make it easier to follow things along.
- Most likely I will add some simple machine learning to try and forcast how things might go. Perhaps I will stick to the more traditional approaches in signal processing, i.e. Fourier Transforms or something along those lines.
- Right now data is written to a flat json file, and for the current purposes that is fine. However, as I transition to an actual frontend interface, I most certainly will need to have some database running so that I can have random access to the data. As it stands I am thinking of a document based database such as Elasticsearch.
- No application like this would be complete without notifications on when a certain crypto has gone above or below a set value. I will be adding that too.
- Maybe, I will add automatic selling tools, but I am weary of any potential bugs that may result in me buying $1000.0 worth of Doge coin (hah, I wish that had happened back in March of 2020).