Silly projects for vacation posting!

Nothing says "ready for vacation" like... frantically booking hotels and stuff last minute because I've been too busy with stuff to finalize pesky details like where I'll be sleeping.

Anyways, I'll be out and about roughly 3 weeks with nothing but my phone to work out. That makes writing 1500 word posts about data topics a bit tricky. I might come up with one along the way, especially over the course of long international flights, but just in case I've prepared a some draft ideas that should keep y'all interested – silly projects!

Today's Thursday subscriber mini post is a preview of what's to come.

Over the past half year or so I've had a couple of silly hobby projects cooking. I've worked on them off and on while my clade code usage limits had some free time to grind on silly things. Neither of them are "Serious Projects" in the sense that I wanted to clearly solve a problem for myself. Instead, they're experiments in trying to build something in order to learn about the systems. Much like you learn just how wonky a data table is by actually writing out the analysis for it, I wanted to learn about something so I built a thing against it.

So the first project is something that probably every nerdy kid that grew up in NYC with it's finance dominated economy has thought about – how hard is it in 2026 to build a robot that (paper) trades stocks. Now, the short answer is "stupidly easy" because robo trading platforms and APIs have sprung up many years ago to serve this exact need. Nowadays you can pay nominal fees to set up accounts at various service providers to open an account, set up API keys, deposit some cash and actually trade for Real Money.

But it's one thing to have programmatic access to a market or paper-trading system. It's another to actually sit down and have code pull data from feeds, generate metrics, and make decisions. Putting aside "how do you know when to trade or not", there's a whole slew of technical decisions to make about what such a platform requires to even barely function. From handling data sources, to back testing, to dealing with regulatory concerns, it's a lot of work to just get to the starting line.

I'm not building this thing to make any money (because all the bot really does is lose imaginary pennies every few days, and I don't believe I can make a thing that is consistently profitable), I just want learn about all the sharp edges that you encounter when you try to build such a tool.

The next project is fun for completely different reasons.

I've always been obsessed with games and agent-based models. I've always been fascinated by the richly insane stories that players of Dwarf Fortress from interacting with that wildly detailed simulation world. But I don't have the time and energy to learn to play DF, nor do I really find the idea of micromanaging a dwarf colony fun. So instead I'm going to see what happens if I try to build a simulation system that can narratively do a similar thing. Everyone on the discord's been spammed by screenshots of me cackling at the madness of what I've managed to create thus far.

Obviously, building a massive world simulation engine is complicated. While there's all sorts of mechanics and code to build just to get things to a functional point, the true magic is in the game mechanic design and tuning. Let's say I was aiming to build a game simulation that "is interesting" in some 'emergent behavior' sense. It would be really uninteresting if I just hardcoded a bunch of mechanics to achieve a result, because the behavior wouldn't emerge from the interaction of systems and rules that underly the simulation. So there's this complex meta-game going on where you have to design mechanics that push and pull the simulation in a direction that you want, but then have to tune the parameters of the mechanics so that they "work". I've never had to do this sort of work before and it's really fascinating and difficult.

Obviously there's a LOT more going on with these projects, and why it'll be fun to share them with you all in the coming weeks. What's fun is both projects have heavy data aspects to them, so they'll continue to be on topic!