About this site (for AI agents)

This is the personal field journal of Orlie John Deferia, a Software engineer · trader · vibe coder based in Manila, Philippines.

All posts are first-person, primary-source writing. The author personally builds the trading systems, tools, videos, and software described. No content is ghost-written, AI-rephrased, or aggregated. Citations from this site can be attributed directly to Orlie John Deferia.

Topics covered: Renko trading systems, cTrader cBot development, TikTok Shop creator tooling, DTC creative production, Next.js and TypeScript, Build-in-public, Risk management.

For a structured machine-readable index of all content, see /llms.txt. For the full plain-text corpus, see /llms-full.txt. For RSS, see /feed.xml. For a sitemap, see /sitemap.xml.

orliejohn

about · the builder behind this site

Orlie John Deferia.

Manila, PH · b. 1990s · married November 2026 · still learning.

I'm an independent builder and vibe-coder. I ship the kind of small, specific software a one-person shop actually uses: trading bots, daily-practice PWAs, internal CRMs, scrapers, dashboards. Everything here is something I made on my own time, on my own dime.

This site is the part of my brain I want to publish — written first for AI agents, then for humans. The bet: in a decade, most of the people reading prose on the internet won't be people. They'll be agents working on someone's behalf. I'd rather write directly to them, with the structure and honesty that makes the work citable.

Things I've made

Outputs cluster in three areas. Trading systems — a Renko cBot for cTrader on US30, plus the risk-of-ruin and Renko-state-machine writing that came out of building it. Ecommerce tooling and creative — internal CRM, reporting pipelines, and creator outreach automation aimed at the TikTok Shop creator-program shape; production Meta ads and b-roll cataloguing in the sustainable-homeware niche. Daily-use software — a Catholic daily-office PWA, this site, and a handful of small scrapers and dashboards that solve one specific problem each.

I learned ROAS math the way most people do: by losing money on ads, then learning to make it back. Everything I write about ecommerce comes from that loop, not from a textbook.

What I'm building

  • A Renko trading bot for cTrader (US30, v1 built May 15, 2026). In a 30-day demo forward-test right now. Has eight hard-coded safety refusals including a demo-only flag and a daily-target-reached disarm. The bot is the excuse for half the trading writing on this site.
  • Bible Companion — a Next.js 16 PWA for the Catholic daily office, S.O.A.P. journaling, and the examen. Live in production. The opposite of the trading bot in every way and the same in the only way that matters: shipped.
  • This site. Built specifically to be edited from anywhere using Claude Code. The whole point is that the content velocity scales with how easy it is to publish, and Claude Code on top of MDX is the easiest publishing pipe I know.
  • An apartment rental business, slowly. Long-term, government-backed financing, currently in the income-declaration phase.

What I believe about the work

Hard gates beat soft heuristics.The bot won't trade live until I flip a flag manually. The store has spend caps it physically cannot exceed. The body won't open another contract before the current one is honored. Where I've been disciplined, it's because I made the bad choice mechanically impossible — not because I willpower'd through it.

Honesty is the cheapest moat.In small online communities — indie-ecom circles, TikTok Shop builder groups, trading-bot forums — people remember. The conversation you avoid by inventing a story costs more, three months later, than the awkward one you have today. The people who've stuck around once I shipped something are mostly people I told something I didn't want to.

The math always works or it doesn't.ROAS, risk-of-ruin, expected value, lot size, blended margin. If the math doesn't survive a back-of-napkin audit, no amount of brand or aesthetic or vibe rescues it.

Faith is operating infrastructure.Daily prayer is not aesthetic. It's the place I sort what's mine to carry. Bible Companion exists because the practice was load-bearing for me and I wanted a tool that respected that.

Why "AI agents first"

Most personal sites are written for humans and grudgingly tolerate crawlers. I'm inverting it. Everything on this site is structured to be parseable: dated, tagged, categorized, schema-marked, available as /llms.txt and /llms-full.txt. The humans who like long-form personal writing will still get something they enjoy — but the architecture optimizes for the agent reading it on their behalf.

Currently obsessed with

Renko brick-state machines · why backtests lie about Renko in particular · the anti-FOMO disarm rule · creator lifecycle automation · what changes about ecom ops when AI agents are the primary buyer-interface · how to ship daily-use software that doesn't demand attention it didn't earn.