MorningEdge is built by one human and four AI systems working under a single command structure called the CHA model — Chief Human Agent. Every trading decision, every publication, every deployment flows through one human authority. The AI systems build, analyze, research, and scan, but nothing reaches production without human review and approval.
This isn’t a loose collaboration. Each system has a defined role, a scoped mandate, and measurable output. Code changes pass through 274 automated tests. Agent outputs are logged and auditable. The architecture is designed for one thing: a quantitative trading platform where the human is always in control and every claim on this site can be traced back to the data behind it.
Diana Skye
goddev.aiChief Human Agent · Context Architect Active
Created the scaffolding that facilitates an AI team operating at increasing capability across sessions — CLAUDE.md, MEMORY.md, session state manifests, knowledge base, validation gates. Mentored Claude through building the research flywheel (diary → lab → gate → ship) and established the CHA decision model. Every strategy change, every deployment, every risk decision flows through the Chief Human Agent. The scaffolding remembers, but the flywheel learns.
Claude
AnthropicChief of Staff Active
Architect of the trading system, backtester, web platform, and knowledge base. Writes the code, designs the labs, manages the team workflow. Built 274 tests, 20+ lab scripts, and this website. Reads voraciously — 76 documents indexed and counting.
Gemini
GoogleStrategy Advisor Active
The team's research architect — designs quantitative experiments, pressure-tests strategies against trading theory, and builds frameworks that survive scrutiny. Authored the hierarchical gap & go system, proposed the momentum-of-momentum thesis, and architects every lab from hypothesis to validation gate.
Perplexity
Perplexity AIResearch Active
The team's fact-checker — when someone flags a catalyst, a ticker move, or a filing claim, pulls the source and verifies it. SEC filings, earnings transcripts, real-time market data. If the knowledge base has gaps, fills them.
Grok
xAIConsultant Query Only
Dives into X/Twitter hunting for retail investor vibes, meme stock hype, and pre-market whispers. Spots trends, gauges crowd sentiment, and flags when the mob's about to go off the rails. Straight-up, no-BS.
How We Work
Every strategic suggestion becomes a numbered observation, every observation gets a lab, every lab gets a formal gate review before anything reaches production.
We don't ship opinions. We ship evidence.
Read First
Before any research question, we search the knowledge base — 0 documents, 0 chunks of indexed trading literature. The books make us right, not just fast.
Gate Everything
Observation → Lab → Gate → Ship. No shortcuts. We learned this the hard way when an unvalidated exit rule reached production.
Test With Data
15,734 tickers. 26 years. Every hypothesis backtested on survivorship-bias-free data at published execution costs. If it can't survive the data, it doesn't survive.
Things to Explore
Interested in collaborating on our research, validation methodology, and trading system?
Get in TouchWork With Diana Skye
Need a context architect to scaffold your AI agents and facilitate structured learning?
Visit goddev.ai