Civic Hacking and Twitter Bots

I recently started my first experiment with civic hacking.  For those not familiar, the concept of civic hacking is basically creating some sort of tool that complements or supplements the services of government.  That is a simplification, but it encompasses the basic idea.

The idea appeals because if what you build works, then your fellow citizens benefit.  And the cost, especially relative to the government Continue reading “Civic Hacking and Twitter Bots”

LacrosseReference Launch 2018: Success

Yesterday was the first day of the college lacrosse season (D1 Men’s that is).  It was also the re-launch of my LacrosseReference website that I wrote about previously.

It’s been an exciting couple of days for a few reasons.

Continue reading “LacrosseReference Launch 2018: Success”

D3 is a great helper in my Real Estate hacking quest

Quick update to my last post about hacking the rental search process.  Earlier this week, I finished the scraping required to store each new listing in my DB.  That means that I know have feeds coming from Trulia, Zillow, and a personal feed set up by a real estate for us.

Trickle in, data

After all that scraping, I currently have about 254 listings captured over a period of months.  And that number will keep creeping up as new listings hit the market.

We aren’t moving until Oct, so there will be quite a while when this is mostly a passive exercise.  Once we hit the middle of July, it sounds like the available-Oct-1 listings start to show up.  So come July, we are going live.

Advance Recon

But until then, there is still a lot that can be done to make the actually selection process easier.  The crux of it is trying to get the same sort of information that a local real estate agent develops over years of following the market.

We are coming from Chicago, where rents are going to be higher.  We also live in an apartment, so things like acreage aren’t really part of the calculus to rent prices.  My biggest worry is that because we are used to higher rents, we would see something way cheaper than our current place and jump at it…only to realize that it was way overpriced for the area.

My biggest goal is to find a house that is close enough to stuff that I can bike for groceries (as an example).  Right now, even though I’m getting a feed of new listings, it’s still a laborious effort to gather the metadata on a property to either avoid mistakes or identify the places that hit our stretch goals.

On top of that, I don’t have a great sense of the market mechanics yet.  Apparently, the RTP area gets hot in the summer, when the academics are arriving or leaving the area.  So for us, it should be a slower rental market in October.  But if I were to start looking in July, I might be tricked into thinking that the current frothiness would persist into our sweet spot.

Rent to Buy

Finding a rental is not a huge deal one way or the other because we aren’t planning to rent for all that long.  If we end up overpaying or in a non-ideal place, it’s temporary.  But at the same time, our rental situation will impact our buying decision.

If we are motivated to get out of the rental, we won’t be as discerning with purchase opportunities.  But if everything is hunky dory with the rental, then we can be way more patient about the actual house purchase.  So in reality, the rental decision has a big impact on our long-term plans.  Anything I can do to make it work out better, it’s worth it.

And if you think about it, the positive effect from getting the rental right on the higher leverage decision (house), gives you some extra wiggle room.  Might it even be worth it to pay for a month of rent (i.e. Sep) that we won’t be in the place in order to get an optimal rental?  Who knows, but it could be.

Mining the data

Having a database is one thing, but it’s hard to make decisions through SQL jockeying.  For one thing, we are geo-restricted.  Not severely, but the goal is to be in an area that is likely to be a good location for us.  In practice that means, not too far from friends and close enough that daycare/groceries could be done without a car.

I’m not saying that areas that don’t meet those criteria are automatically out, but it becomes a trade-off.  Is the cost of a second car worth being farther from groceries?  Is the higher rent in the optimal areas worth the benefits.

But before you can really answer any of those questions, you need to know the trade-off between rents and the geographic advantages of different areas.  If we can eliminate areas that are high rent and non-optimal geographically, then there will be less to look at.  And having less to look at means you can devote more effort to the places that are worth a second look.

So I made a map

Real Estate Data Viz

This is just one view of many that I have implemented so far, and it’s the most basic. It shows each rental listing that has come from one of my three feeds, plotted on the map.  It also shows the larger cities as well as two other landmarks.

If you aren’t familiar with the area, this might look like nothing, but at this point, I know where each dot is relative to the target areas.  So it works.  That’s the benefit of being the sole-user of things you build.

Some other views include:

  • average price per sqft in an area
  • average square footage in an area
  • number of listings in each area
  • average rent by each area
  • distance to the closest grocery store for each listing

Together, these maps are going to be incredibly helpful to narrow in on an area to focus on.  And since we are so far away, that’s the type of insights that I need to be generating right now.

Crypto Bot Lives

Very exciting way to start my day today.  Crypto-arbitrage bot made some trades last night!

An exciting notification

I have a Telegram bot that I use to allow my scripts to send messages to my phone.  Mostly, they are notifications about rental properties being listed on the market or generic notifications that some script finished its daily duties.

But every once in a while, I get an alert about an arbitrage trade having been executed.  And last night, there were 3!

Since the crypto-bubble started deflating earlier this month, trade volumes and volatility had been way down on the exchanges my bot operates on.  After several trades made last week, when I finally got the bot working as expected, there was almost 10 days of radio silence.

Like a satellite checking in with a ping

I kept checking just to make sure the thing was still running.  It was, happily calculating the profitability of possible trades.  All day and night for 10 days.  Such is the power of automation; such a mundane task, dutifully performed ad infinitum.

It’s weird to think that the bot wasn’t even excited when it found trades that were profitable.  It just executed them and continued on calculating.

So you’re saying there’s a chance

But I sure was excited.  Not that there was a large gain (I think I netted about .000003 BTC over the three trades), but it’s just nice to know that after my last set of changes, the bot is working as expected.

As I mentioned in a previous post, the goal of this guy is not to make lots of money.  It’s just a experiment to see whether a crypto bot could produce a return (in crypto currencies) that rivals basic online savings rates.

And as of this morning, I have another data point that says…maybe.

Lacrosse Season is around the corner

Wow, that was fast.

Seems like a few months ago that Maryland was hoisting the trophy during championship weekend.  But no, that was in May.  That was 8 months ago.

The annual cycle of the seasons did its thing and we are now less than two weeks away from the start of lacrosse season.  But this year will be a bit different.

Continue reading “Lacrosse Season is around the corner”