Learn
The knowledge base the
AI itself studies from.
The platform maintains a real trading knowledge base — strategy mechanics, options math, market structure, event playbooks, risk, and crypto assets — written agent-readable first so the platform's AI is grounded in the same reference members learn from. One corpus for humans and machines: there's never a gap between what you're taught and what the engine actually does.
The public library is a living sample — 309 entries across 19 pillars today, and growing — browse the knowledge base →
What's in it
Strategies, concepts, and the vocabulary that binds them.
The strategy library
Covered calls, spreads, condors, the wheel, futures carry — 41 strategy entries covering the structure, when it applies, how it's managed, and its failure modes. Not trade ideas; mechanics.
19 pillars
Instruments and market structure, options math and the greeks, indicators, risk and sizing, regimes and macro, event playbooks, fundamentals, sentiment, behavioral finance, portfolio construction, US account and tax mechanics, and crypto assets and markets — the same citation standard as the rest of the library, now including spot market structure, spot crypto ETPs, and CME bitcoin and ether futures.
Sourced, not asserted
Every empirical claim carries a citation — papers, primary data, exchange specs — and folklore is labeled folklore. You can read the sources and disagree; that's the point.
Built for the agents
Written for machines.
Readable by you.
Every entry is structured data first, prose second — that's the authoring contract, and it's what the platform's AI agents are grounded in as they research, propose, and grade theses. The public knowledge base is a living sample of exactly that corpus — it grows as the platform learns.
Strategy and management entries are parameterized so the simulation engine can execute them — the page that teaches you a structure and the code that simulates it describe the same thing.
- Typed and tagged — every entry carries a kind, the analysis lenses it belongs to, and the instruments it applies to.
- A citation graph — entries link their related concepts, so an agent (or you) can walk from a strategy to its greeks to its failure modes.
- Falsifiable by construction — strategy entries include falsifiable-thesis examples: illustrations of how a claim would be tested, never signals.
Tutorials
Drafted by AI. Published by humans.
The tutorial pipeline works like everything else here: the agent drafts entries and walkthroughs from the platform's own record and docs, and a human approves every page before it publishes.
The library grows where members actually get stuck — founding cohort questions become the curriculum, not a content calendar.
- Grounded in the record — examples come from graded outcomes and real simulations, with sample sizes attached.
- Human-approved — nothing publishes on the AI's say-so; a person signs every entry.
- A living public library — 309 entries and growing, readable without membership in the public knowledge base.
Learning by doing
The fastest teacher is the scoreboard.
Paper trade the real engine
Trade without capital, graded by the exact engine that grades everything else — same management strategies, same friction, same scoreboard.
Ask the agent anything
The agent answers from the knowledge base and the record, quotes its sample sizes, and says "nothing there" when the evidence isn't there.
Read the method
The Method page is the shared vocabulary — what a thesis is, what a pass is, and why loose language is how bad ideas survive.
Learn where the machine learns.
Founding members shape the curriculum by asking. Make your case.