You are the verdict
Judex can do everything except the one thing that matters. That one is yours — by design, not by limitation.
Every ask ends at the same screen: the action, the size, the reason, the price the engine saw. From there, exactly three things can happen — and all three are first-class outcomes. There is no wrong button. There is only a recorded one.
Three verdicts, all of them graded
| Your verdict | What happens | How it's graded |
|---|---|---|
| Approve | executes on the agent book; on the main book it becomes a commitment you fulfil by hand | from your fill forward — execution gap included |
| Reject | nothing trades; the un-acted signal may return tomorrow | on what the recommendation did after you said no — a dodged loser counts for you |
| Let it pass | the link expires quietly; nothing trades | like a rejection — but the scorecard remembers it as silence, not as a decision |
There's a fourth state you can reach only on the main account: approved, but never traded. You said yes in the app and the trade never appeared at the broker. The ledger holds it apart from both agreement and refusal — because it is neither.
Every ask carries two grades — the recommendation's and yours
One row, two questions. Was the system right? — judged from the suggested price forward, as if the recommendation had been followed exactly. Did your verdict favour you? — judged from what you actually did. The two are independent, and they often disagree.
A buy call is proved right by a rise. A sell call is proved right by a fall. If you approved, your grade moves with the recommendation's. If you skipped, yours is the recommendation's mirror image — skipping a right recommendation cost you, skipping a wrong one saved you.
| What the price did next | You approved | You skipped |
|---|---|---|
| proved the recommendation right — the buy rose, or the sell fell | system ✓ · you ✓ | system ✓ · you ✗ |
| proved the recommendation wrong — the buy fell, or the sell rose | system ✗ · you ✗ | system ✗ · you ✓ |
| barely moved yet | too early to grade — both grades wait | |
The third row exists on purpose. A call made this morning is still priced off its own bar; stamping it right or wrong would be noise. Until the price leaves a narrow band around the suggestion, the row reads too early to grade — and it sits out of the hit rate entirely, on both sides of the ratio. Every grade carries "so far": it moves with the price until the position closes.
Two accounts, two meanings of "approve"
On the agent account, approve is the trade: the order goes to eToro within moments, the fill comes back, done. On the main account, Judex never holds the pen — approve records your intent, and the trade is yours to place. Reminders follow at +10 minutes, +1 hour, +2 hours; the reconciler then watches the broker for the trade itself, however and whenever you place it.
That watching is generous: it matches your real fills — even ones better or worse than suggested, even hours late, even trades nobody asked for. What you actually did always outranks what was proposed.
What your verdicts build
This is the second pillar doing its work. Month by month, your verdicts accumulate into the behaviour side of the Scorecard: how often you follow the system, what your execution gap costs or earns, whether your rejections beat the recommendations you refused — and what the asks you ignored would have been worth.
None of it is a judgment. All of it is a mirror. The point isn't to make you obedient — it's to show you, with numbers, which of your instincts deserve more trust than the system, and which deserve less.
Read on
The glossary → defines every word the asks and the scorecard use — including each verdict state you just met.