I am cheap, and don’t want to pay for hosting, so instead I chose to build my site in Circleci and have CircleCI push the static files up to Github. To do this, I needed to create a branch (or it could be its own repo) to store the static files generated by Jekyll. I chose to use master for static files, and develop for all the Jekyll code because github’s user-pages must be on the master branch.

The first step is just to get your site building in CircleCI.

# .circleci/config.yml
version: 2.0

jobs:
  build:
    docker:
      - image: circleci/ruby:2.4.3
    steps:
      - checkout
      - restore_cache:
          keys:
            - v1-bundle-\{\{ checksum "Gemfile.lock" \}\}
      - run: bundle install --path vendor/bundle
      - save_cache:
          key: v1-bundle-\{\{ checksum "Gemfile.lock" \}\}
          paths:
            - vendor/bundle
      - run:
          name: Install Prerequisites
          command: "sudo apt-get update; sudo apt-get install -y rsync"
      - run: 
          name: Build Site
          command: JEKYLL_ENV=production bundle exec jekyll build
      - store_artifacts:
          path: _site
          destination: /
      - run:
          name: Deploy
          command: _scripts/deploy.sh

This simple CircleCI 2.0 config will build a Jekyll site in ruby 2.4.3, store the static files as build artifacts, cache the installed gems to speed up future builds, and finally run a script to deploy it.

#!/bin/bash
# _scripts/deploy.sh

DEPLOY_DIR=$(mktemp -d)
BUILD_DIR=$(readlink -f "./_site")

# clone target repo to a temp dir to ensure no name collisions
git clone $CIRCLE_REPOSITORY_URL $DEPLOY_DIR
cd $DEPLOY_DIR

# verify clean repo
git fetch origin
git checkout master

# configure git committer
git config --global user.name CI
git config --global user.email ci@example.com

#copy compiled site from artifacts
rsync -a --delete --exclude=.git $BUILD_DIR/ .

# commit changes
git add -A
git commit -m "Deployed in CI: $CIRCLE_BUILD_URL"
git push origin master

# ping search engines with new sitemap
curl "http://www.google.com/webmasters/sitemaps/ping?sitemap=https://www.example.com/sitemap.xml"

The script simply clones the repo, checks out the master branch, syncs it with the output from the build step, and commits it with a link to the build on CircleCI.

After creating a checkout key in CircleCI with write access to your repo, you are now free to use any gems you like and Github will host it for free.

As an extra bonus you can add extra features like spell check, HTML validation, or a safety check to verify that you don’t accidentally add commits to the wrong branch.

As it happens, I too am “in ops”. I choose to introduce myself that way, likely for the same reason as everyone else; because my real job title doesn’t come close to describing what I do, and all the alternatives either don’t apply or don’t make grammatical sense.

Officially my job title is Senior Linux Administrator, but for me that brings to mind a bunch of scruffy neckbeards artisanally hand-crafting config files and shunning daylight. Instead, I spend most of my days automating installs and config changes with Chef and Etch. I am also singularly responsible for our network infrastructure and firewalls (Network Engineer/Administrator). I often need to write software to help test and troubleshoot the systems I’m building, and occasionally even write the software the finished systems will run (Software Developer). I spend a sizable amount of my time setting up, optimizing, and reacting to alerts in our various monitoring systems (Site Reliability Engineer). And when I’m not doing any of those things I’m learning new programming/markup/query languages trying to find the slow spots in our infrastructure so we can save a few milliseconds on each web request (Data Scientist).

A few years ago the term DevOps started getting popular as a way to describe how Developers and Operations can work together using the same tools to be more efficient and build better products. Some people have tried to use it as a job title such as DevOps Engineer, or DevOps Administrator. The problem with these sorts of titles is they just don’t make sense, you might as well call yourself an Agile Engineer, or a Teamwork Administrator.

Operations Engineer isn’t bad, but that’s only one step away from “I’m in ops”, and anyone outside the tech field would likely think it has something to do with trains. I guess for now I’ll just have to keep telling people “I’m in ops”, at least until I find a better title I can steal.

To dig into this a little deeper, we need to start with some assumptions

1. The home we buy must be of equal or greater value to the home we sell at all times.

   We’ll be looking at houses in the same city which should be rising and falling at the same rate, and we want to make sure we’re not downgrading.

2. Rent will not be less than the mortgage payment.

   Landlords usually have mortgages too, and we don’t want to downgrade too much during the rental period.

Looking at home values over the last few years, you can clearly see a seasonal rise and fall in home values, higher in the summer and lower in the winter. This gives hope that it might really be possible to make money by delaying buying after selling.

We want the housing market to lose as much of its value as possible between selling and buying 6 months later. To make any profit, home values must drop enough to cover selling costs, the increase in monthly housing costs, and the principal we would have put into our house during the rental period if we hadn’t sold.

cost_of_renting = monthly_rent * months
cost_of_owning  = (monthly_mortgage - monthly_principal_paid) * months

revenue         = sell_price + cost_of_owning
expenses        = buy_price + selling_costs + cost_of_renting

# profit        = revenue - expenses
profit          = (sell_price + monthly_mortgage*months) - (buy_price + selling_costs + monthly_rent*months - monthly_principal_paid*months)

Based on my completely unscientific practice of randomly looking at different areas in Trulia, it seems like summer time generally accounts for almost all of the yearly growth, and the winter generally loses about 10-20% of what the summer gained. Since we’re only concerned with losses, and since year-over-year growth in my area last year was are around 20%, I’m going to use a buy_price 4% less than my sell_price (20% * 20% = 4%). To help our profits a little more, I’ll also assume monthly_rent and monthly_mortgage are equal. Finally, selling a house in a reasonable amount of time usually requires a realtor, and realtor commissions are usually about 6%, so we’ll stick with that.

sell_price             = 300_000
buy_price              = 300_000 * (1 - 0.04) # 288,000
selling_costs          = 300_000 * 0.06       # 18,000

monthly_rent           = 1_500
monthly_mortgage       = 1_500
monthly_principal_paid = 300

cost_of_renting        = 1_500*6              # 9,000
cost_of_owning         = (1_500 - 300)*6      # 7,200

profit = (300_000 + 7_200) - (288_000 + 18_000 + 9_000) # -7,800

That puts us $7,800 in the red for selling our hypothetical house in the summer and buying one of equal value in the winter. But there is still the income from selling the house to consider. Assuming 25% equity, that gives us $57,000 to invest elsewhere while we rent.

selling_costs = 300_000 * 0.06         # 18,000
equity        = 300_000 * 0.25         # 75,000
revenue       = equity - selling_costs # 57,000

Historically the stock market as a whole averages 10.4% per year, and if we could count on that return we would make $2,890.70 in 6 months, though the portfolio required to get that rate would be almost as likely to lose that much in the 6 months we could invest it. The better choice would be to invest the majority of that money in bonds, CDs, or money markets that are much safer but unfortunately have much lower returns (0.5-1% currently), which only gets us $284 in 6 months (not counting brokerage fees).

future_value = present_value * (1 + rate)^time
# risky portfolio
57_000 * (1 + .104)^.5 # 59,890.70

# safer portfolio
57_000 * (1 + .01)^.5 # 57,284.29

So, in a perfect world where we can sell at the absolute top of the market and buy at the absolute bottom 6 months later, and find a suitable rental between for the same price, the total profit (not counting moving expenses, storage, taxes, brokerage fees, closing costs, title insurance, HOA fees, or anything else I’ve forgotten) is $-7,515.71.

The biggest expense cutting into our profits from this example was the realtor fees, as you can see from the graph below, as long as the market drop is less than the realtor fees, its almost impossible to come out ahead. While there are ways to save money on realtor fees such as selling yourself or using an online listing agent, short-selling houses just doesn’t seem like a practical way to make money to me.