About
Claims are cheap.
Proof is the product.
Trading content is an ocean of claims nobody has to prove — signals with no track record, gurus grading their own homework, screenshots of one good week. We built the place where every idea — the machine's, the community's, yours — faces the same discipline: declare it, make the evidence immutable, validate it forward, grade the outcome.
Principles
The rules the platform runs on.
Backtests propose. Forward evidence disposes.
Any slice of history can be made to look like an edge. A thesis here is judged on data that did not exist when it was declared — new simulated cohorts first, real fills last. Nothing is trusted for scoring well on the past it was mined from.
Evidence is immutable from declaration.
The moment you declare a thesis, the case as it read right then — the backtest slice, the regime, the market context — becomes a permanent, immutable part of it. You can never fool yourself later about what you knew.
A pass is a decision, and it gets graded.
Declining a signal records a reason, and the simulator keeps tracking the trade you didn't take. The discretion scoreboard tells you, reason by reason, whether your no's make money or cost it. Nobody else grades restraint.
Disproven ideas stay on record.
An idea that failed forward validation is kept, with its evidence, precisely so it can't sneak back in six months later wearing a new name.
Simulated numbers earn trust against real fills.
Every real trade is audited against its simulated twin. Until the simulator tracks reality within friction noise, its numbers justify analysis — never position size.
Lines we don't cross
What we refuse to do.
- Trade without your mandate. The AI researches, grades, and drafts — and every order runs on authority you granted. Automation is opt-in, bounded by rules you wrote, and never the default.
- Sell signal noise. A quiet dashboard is the system working. We will never manufacture trade frequency to feel useful.
- Claim an edge before the record shows one. Positions without a forward-validated thesis are research, and the platform says so — instead of inventing conviction.
How we support
Feedback becomes shipped work.
Usually within weeks.
Most platforms treat feedback as sentiment — something to survey, thank you for, and forget. Here it's a work order. We run the company the way we run the book: every claim logged, every priority explicit, every outcome checked.
Logged, never lost
Every report becomes a numbered case the moment it arrives — your words, the exact build you were on, the context around it. There is no suggestion box here; there's a ledger.
Prioritized honestly
Cases are triaged against one explicit priority order — not whoever asked loudest. When something's deferred, it's deferred on the record, with a reason.
Shipped in weeks, not quarters
Turnaround is measured in days and weeks. Members routinely see their report show up as a shipped change before they've stopped expecting a canned reply.
We can sustain this because we built the machinery for it — the same discipline that grades trades grades our own roadmap, and it gives a focused team the operating leverage of one many times its size. Reply to any email from us and it reaches the people who build, not a queue.
The company
Thesis Trading AI
Built by a trader who wanted his own decisions graded honestly, and opened to the people who want the same. The platform and this site are operated by Thesis Trading AI LLC, a US company — the entity behind the Terms of Service and Privacy Policy.
Membership is by inquiry or invitation.
Founding membership is assembled deliberately, not first-come-first-served. Tell us how you trade, and make your case.