/ JavaScript

Run a Crypto Trading Bot on Cloud Foundry

Everybody is talking about Bitcoin and its peer currencies. The hype is so great that even small investors, who have no idea about financial constructs, like myself want to get aboard the investing and trading train. That's why I recently decided to create my own account on GDAX and try my luck.

Trading, at first, was fun but I realized pretty quickly that I'm not the kind of person who wants to watch their dashboard every five minutes. Furthermore, I have no clear strategy about when to invest in what and when to actually pull out. Being able to read the price charts correctly requires a lot of training and knowledge. However, there are patterns which can be recognized. Since this is a purely analytical work, I figured that it would be the perfect job for a computer. So I looked online and found my answer with Zenbot. It's an open source, cloud ready trading bot for crypto currencies. Let's deploy it to Cloud Foundry.

Important: The presented software and concepts are still very young. Please use them with great caution and don't start out with investing big amounts of money right away.

Clone Repo

First, we need to clone the repo from GitHub. It's advisable to check out the latest tag so you're not trusting all your money to a possibly unstable master branch.

Create Services

Zenbot has two dependencies it needs to work. The first one is MongoDB. In most Cloud Foundry installations, you can get that with a simple command like:

$ cf create-service mongodb small zenbot-db

The next dependency is a GDAX API client. Any of the big trading platforms will do but I'm using GDAX to keep things simple. Here you can find a list of the supported platforms. To use GDAX, visit their website and create an account. The process is a little tedious because you'll have to provide a photo ID and go through a couple of more verification steps. But, believe me, eventually it's worth it. After you are signed up, choose the product you'll want to trade (e.g. BTC/EUR) in the top left corner and transfer some money to your account (either from a credit card, from a bank account or from an existing Coinbase account.

After that's done, go to the API settings and create a new set of API keys with the "Trade" permission. Note down the credentials and create a user provided service for them (be aware that this will allow anybody with the respective permissions in Cloud Foundry to see your API credentials):

$ cf create-user-provided-service zenbot-gdax -p '{"key":"YOUR-API-KEY","b64secret":"YOUR-API-SECRET","passphrase":"YOUR-API-PASSPHRASE"}'

In case you are using a different trading platform, create a similar user provided service with their API credentials.

Configure the Bot

Now that we have the services set up, we'll need to tell Zenbot to use them. Copy conf-sample.js to conf.js and add the following lines to automatically read the respective config items from the VCAP_SERVICES environment variable Cloud Foundry exposes:

// Add this part after the existing MongoDB configuration

if (process.env.VCAP_SERVICES) {
  const creds = JSON.parse(process.env.VCAP_SERVICES).mongodb[0].credentials
  c.mongo.host = creds.host
  c.mongo.port = creds.port
  c.mongo.db = creds.database
  c.mongo.username = creds.username
  c.mongo.password = creds.password
}
// Add this part after the existing GDAX configuration

if (process.env.VCAP_SERVICES) {
  const creds = JSON.parse(process.env.VCAP_SERVICES)['user-provided'][0].credentials
  c.gdax.key = creds.key
  c.gdax.b64secret = creds.b64secret
  c.gdax.passphrase = creds.passphrase
}

This will ensure that the config is read automatically from the environment, if the VCAP_SERVICES variable is there.

Furthermore, we'll make it possible to configure the traded product and the strategy to use through environment variables. For that, chage the following lines:

// default selector. only used if omitting [selector] argument from a command.
c.selector = 'gdax.BTC-USD'
// name of default trade strategy
c.strategy = 'trend_ema'

to these

// default selector. only used if omitting [selector] argument from a command.
c.selector = process.env.ZENBOT_SELECTOR || 'gdax.BTC-USD'
// name of default trade strategy
c.strategy = process.env.ZENBOT_STRATEGY || 'trend_ema'

This allows us to set the selector (i.e. the product) and the trading strategy the bot should use through environment variables. To get a list of possible values, you can run the following commands in the root folder of the repo:

$ ./zenbot.sh list-selectors
$ ./zenbot.sh list-strategies

Create manifest

The last step before pushing your bot is to create a manifest.yml file to automate the setup of your deployment. It should look as follows:

---
applications:
- name: zenbot
  memory: 256M
  buildpack: https://github.com/cloudfoundry/nodejs-buildpack.git
  no-route: true
  health-check-type: process
  
  services:
  - zenbot-db
  - zenbot-gdax
  
  env:
    ZENBOT_SELECTOR: 'gdax.BTC-EUR'

Replace the selector with the respective product you want to trade and add another line if you don't want to use the default trading strategy.

Push

Now, all that's left is to run cf push to deploy your bot. After a successful deployment, you can use the following command to get a live log stream to see exactly what your bot is trading:

$ cf logs zenbot --recent

See Reading the Console Output for instructions about how to interpret these logs. Esentially, if the last number is green, you should be happy with the bot and if the second to last number is green, you should be happy in general ;-)