Skip to Main Content

Desire-By-Design

Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

Demo video · github.com/…

Project name: InsideTrader's One line: The trading coach that runs real paper accounts in public — so a beginner learns how trading actually works by watching real trades, placed under strict rules, explained in plain words, without risking a dollar of real money. Live product (public record page): https://muse.ai/s/insidetrader-s-live-xbxy69xvxamixd5h/DWdxGhmBWOb61H4zUz7_vbEoQYEqaZWrNr6h1aKrRc0 What it is. InsideTrader's is a paper-trading education product, live and operating today — not a prototype built for this event. It runs four paper-trading accounts ("lanes"), each funded with $30,000 in paper capital: a day-trading account (opened Sep 30, 2026), a legacy/main account, a commodity account (opened Oct 5, 2026), and a crypto account (opened Oct 6, 2026). Every lane trades under rules fixed in advance: every position carries its exit before it opens, position sizes are capped, winners are protected by trailing exits, and the day-trading account is flat by the market close every single day. The trades are mirrored to an Alpaca paper account and reconciled — on Nov 2, 2026 the reconcile was clean in all four lanes, no drift. Who it is for — the beneficiary, named. The ordinary person who wants to learn to trade and would otherwise learn the way most beginners learn: by losing real money in their first year. InsideTrader's exists for that person. They watch a full trading operation run in the open — a written morning board before the market opens, trades placed only when the stated conditions print, every result published, green or red — and they learn the discipline before they ever risk rent money. The founder knows this problem firsthand: he taught himself the market from newspapers and the news with no capital to spare, and built the product he needed then. That is Build for Your World, literally: a problem the builder lived, solved in his own field. The operating record (all paper, all published daily). A market briefing and trading board have been produced every day from Sep 24 through Nov 2, 2026 — 40 consecutive daily editions, archived. Headline results as of the Nov 2, 2026 close: - Day-trading lane: $30,000 → $30,798.58 since Sep 30, 2026. - Most recent trading day (Mon Nov 2, 2026): +$118.36 banked, green — five trades, every one closed the same day: IDXX +$7.26, INO +$138.60, RIVN −$27.50, CROX $0.00, INO re-entry $0.00. - Legacy lane: +$1,242.35 cumulative realized, currently flat. - Commodity lane: $29,831.26 — cumulative banked −$168.74. - Crypto lane: $30,000.00, banked $0.00; its one armed trade has never triggered. We publish the losing lane and the lane that never traded in the same table as the winners, because that is the actual lesson: most of trading is discipline, small losses, and trades you correctly never take. How Meta's models are integrated. InsideTrader's integrates the Meta Model API through a registered connector built for this product, verified live on Oct 23, 2026 with real calls: the models endpoint returned HTTP 200 listing muse-spark-1.3, muse-spark-1.2, muse-spark-1.1, muse-image-1.0, muse-voice-transcribe-1.0, sam-3.1, and muse-wonder-1.0, and a chat completion against muse-spark-1.3 returned HTTP 200. In the product architecture, Muse Spark is the teaching layer: it does the volume work of turning a completed trade — the setup, the entry, the exit, the result — into a plain-words lesson a beginner can actually absorb, and carries the analysis workload across the full archive of trading days held in its long context. The rules engine decides; Spark explains. The machine never talks a learner into a trade — it teaches them what already happened and why the rules did what they did. One honest limitation, stated plainly. The Meta-model integration is verified at the API level and wired into the product's teaching layer, but it is not yet the always-on tutor inside every screen of the app — today, the daily operation (board, trades, lessons) is produced by the product's own system, with the Spark layer verified and integrated rather than load-bearing on every trade. And every result above is paper money: that is the point of the product, but it means the record proves discipline and teachability, not real-money performance. We would rather submit both facts in writing than have a judge discover either one.

Team