Seven stages.
Any one can end it.
A setup does not become a signal by being interesting. It becomes one by surviving, in order, every stage below. Most do not, and each one that fails keeps the reason it failed.
The seven stages
In order, with nothing skipped. Scroll to follow one setup through.
- 1
No AI spent here
Screen everything
Every eligible spot and perpetual pair is ranked by liquidity, spread and volatility from public exchange feeds, continuously.
- 2
Closed candles only
Qualify on structure
A setup needs causal market structure and strategy-specific indicator confluence before it is even called Potential.
- 3
In parallel
Seven independent analysts
Structure, technicals, liquidity and derivatives, dated news, base-rate statistics, market regime, and an adversary paid to argue the other side.
- 4
Grounded or rejected
Evidence check
Every number an analyst writes must match computed evidence. Missing core research or a material contradiction ends the setup.
- 5
DeepSeek, second to last
Chief analyst
Reads every finding and makes the desk's call with a conviction score. Below the admin-set threshold, the setup never reaches the final judge.
- 6
Always the last word
JEV System One
An external judgment service weighs the whole package, including the chief's synthesis. No explicit approval, no signal.
- 7
Can only veto
Risk gate
Deterministic checks on price freshness, spread, slippage, R:R after costs and open exposure. It never approves anything on its own.
Four rules the desk does not bend
- Closed candles only
- Structure, indicators and patterns are computed after a candle closes, so a result cannot repaint.
- Numbers must be grounded
- Every figure an analyst writes is checked against computed evidence. A mismatch rejects the package.
- One analyst argues against
- The adversary exists to find the strongest case against the trade, and that case is published with the signal.
- Risk can only veto
- The risk gate never approves anything on its own. It can stop a signal; it cannot create one.
Nine ways a setup ends without a signal
Declining is the normal outcome, so the desk names each kind. Open any closed setup in the app and you will see which one it was, who decided, and what each analyst said.
Price moved before approval
The market left the entry zone while the review was still running. Chasing it would change the trade.
Market data went stale
Prices or candles were too old to trust, so the review stopped rather than guess.
Analysis could not be completed
Core research was missing. A partial case is not a case.
Evidence was not strong enough
The numbers were real but did not add up to a trade worth taking.
Analyses disagreed
The analysts contradicted each other on something that matters.
Did not pass the final review
The chief analyst's conviction was too low, or JEV System One did not explicitly approve.
Outside risk limits
Spread, slippage, reward-to-risk after costs or open exposure failed a deterministic check.
Expired before review finished
Every setup has a clock. When it runs out, the setup is closed.
Review interrupted
Something failed on our side. When that happens nothing is published.
The record of one review
This is what a closed setup looks like in the app: each analyst's stance and confidence, the consensus, the four checks that follow, and the outcome. The example ends the way most do.
When a setup does pass, the signal carries an entry zone, a structural stop, up to three targets with reward-to-risk for each, a risk label with its reasons, an expiry and the written analysis. See what reaches you
- StructureSupportsHigher low held above the 4h demand zone.
- TechnicalsSupportsRSI back above 50, MACD crossed up.
- Liquidity and derivativesOpposesFunding is rich and longs are crowded.
- NewsNeutralNothing dated in the last 48 hours.
- StatisticsSupportsBase rate is above average in this regime.
- Market regimeSupportsTrending, with volatility contracting.
- AdversaryOpposesA sweep of the range low breaks the idea.
- 4 of 7analysts support the trade
- Evidence checkEvery figure matches the computed value
- Chief analystConviction 71, the desk needs 65
- JEV System OneApproved
- Risk gateReward-to-risk after costs is under the minimum
Not published. The risk gate vetoed it, and the reason stays on the record for anyone who opens the setup.