< Back to Blog

They Forced Me Back to the Office. My Bot Didn't Notice.

TL;DR / Key Takeaways

  • Return-to-office mandates have no effect on distributed automated systems, which is the point.
  • The gap between "productive" and "visibly productive" is where corporate culture extracts most of its tax on your time.
  • An automated trading bot running on a home server earns the same whether you are in an office, on a beach, or asleep.
  • Location independence is not a perk. For algorithmic systems, it is the default state.

A few months ago, the company issued an updated policy. Hybrid work arrangement. Come in a minimum number of days per week. Specific days may vary by team and manager.

I have been doing this long enough to know what this means in practice. It means: be visible. Show up. Occupy the desk. Participate in the ambient social fabric that makes management feel like the organization is functioning.

I do not say this to be bitter. I say it because after 30 years in software, I have developed an accurate read on what decisions are actually about versus what they are framed as being about. The RTO mandate is framed as collaboration and culture. What it is about is the anxiety of managers who cannot evaluate output directly and so use physical presence as a proxy.

I put on the noise-canceling headphones within about 20 minutes of getting to my desk. Everyone does. The open-plan floor is too loud to think in. So we all sit three feet from each other, headphones on, staring at screens, doing exactly what we were doing at home. Except now we commuted.

The Bot Stayed Home

Here is what happened on my first mandatory RTO day: nothing.

My Kalshi weather trading bot ran its GFS ensemble scan at 06:00 as scheduled. It pulled the latest Open-Meteo forecast data, scored all available temperature contracts across the four-factor system, found two trades with sufficient edge above the minimum confidence threshold of 0.30 and minimum ensemble edge of 0.10, and placed both orders via the Kalshi API before I had finished my first cup of coffee.

I was in the parking garage at the time.

The bot does not know where I am. It never has. It runs on a headless Ubuntu VM on a server in my home office in Atlanta. It has no concept of commute. No concept of whether I am "in" or "out." It checks the GFS ensemble data, scores the markets, applies the filters, and executes. That is the entire loop. My physical location is a completely irrelevant variable.

This is not a coincidence. When I designed the system, location independence was not a feature I added. It is just what you get when you build an automated system correctly. The bot is not location-dependent because there is no logical reason for it to be.

The commute did not make the bot faster. The commute did not improve trade quality. The commute did not help the 62 GFS ensemble members converge on better temperature forecasts. It just cost me 90 minutes of my day and a tank of gas.

What "Passive" Actually Means

People hear "passive income" and picture someone lying on a beach while money appears in their account. That framing does the concept a disservice.

What passive income actually means, at least in the context of automated trading, is that the income-generating activity is decoupled from your physical presence and moment-to-moment attention. The work happened earlier, when you designed the system, wrote the code, tested the filters, and debugged the edge cases. The ongoing operation does not require you to be watching.

My Alpaca stock bot runs on the same server. It monitors technical indicators and Finnhub sentiment data, feeds signals into the Kronos AI model for time-series predictions, and executes trades during market hours. It ran 14 cycles while I was in the car driving to the office.

I checked the logs on my lunch break from the break room at work. Everything was in order.

This is the part that crystallizes something I already knew intellectually but felt more sharply sitting in that open-plan office: the income generated by those systems has no relationship to whether I performed my presence correctly. It does not care about my commute. It does not care about whether I was in the building. It is a function of the quality of the models, the discipline of the filters, and the reliability of the infrastructure. All of which I built. None of which requires me to be visible.

The Real Problem with RTO

The RTO debate in corporate America tends to focus on productivity. Studies get cited. Executives push back. Remote workers argue their output numbers. The conversation goes in circles because both sides are arguing about productivity when the actual issue is something different.

The actual issue is control.

In an office, if your manager walks past your desk at 2 PM and you are at your keyboard, the manager sees a productive employee. If you are working from home at 2 PM and your manager cannot see you, the manager has to actually look at your output to know whether you are productive. For managers who built their entire evaluation model around visual proximity, this is genuinely uncomfortable. The RTO mandate resolves that discomfort by restoring the ability to use presence as a proxy.

This is not malicious. It is just a system optimizing for the wrong thing.

Automated trading systems cannot optimize for the wrong thing. If the Kalshi bot's confidence threshold is set wrong, the bot takes bad trades. The database does not record "manager felt reassured." It records win rate, P&L, and whether the ensemble edge held up. There is no proxy available. The system either performs or it does not.

That directness is worth something. Working in an environment where performance is genuinely measurable, where the scorecard is the actual log file and not someone's impression of your visible effort, changes how you think about your work.

What I Am Building Toward

I am 60. I plan to retire in the Philippines in the next five years. My wife is Filipina and we have been planning this for a while. The dream is a quiet life on a modest budget in a place where the cost of living is a fraction of Atlanta.

For that to work, I need income that does not require me to be in any particular building at any particular time.

The automated trading systems are part of that plan. Not the whole plan, but a piece of it. The Kalshi weather bot and the Alpaca stock bot together represent a model of income generation that is genuinely location-independent in a way that my day job is not. The day job is getting less location-independent by the quarter.

The gap between those two realities is what drives the work I do at 5 AM before I have to get in the car.

If you are a developer or data engineer who has had the same thought, the Predict & Profit bot source code is at predictandprofit.gumroad.com. It is not a passive income magic trick. It is a Python system that uses real public data to find real market inefficiencies and trade them automatically. Building it and running it requires actual engineering skill. But if you have that skill and you are tired of trading it for a seat in an open-plan office, it is worth understanding how these systems work.

The bot is running right now. It does not know I wrote this.