The process, start to finish

How we work, hour by hour

Everything below runs on the server on New York time, with your own local time beside it. Each phase says what the desk reads, why it matters and the rule it produces.

  1. PHASE 0108:30 New York (--:-- your time)

    Pre-market screening & macro bias

    US 10-year yield, macro calendar and a screen of the fixed universe of 25 liquid high-beta names. If the data does not pass the filters, the desk stays flat.

    Read the logic of this phase
    What it observes
    Rates, scheduled releases, live volume, spreads and participation across the fixed universe.
    What it decides
    How the session will be judged before an order exists.
    What it refuses
    Stale prices, silent venues and names outside the approved universe.
    Why it protects
    No incomplete picture is allowed to become financial exposure.
  2. PHASE 0209:30–09:45 New York (--:--–--:-- your time)

    Protective no-fly zone

    Fifteen minutes of pure observation at the bell. Wide spreads and algorithmic fake-outs are filtered by waiting for the first 15-minute candle to close.

    Read the logic of this phase
    What it observes
    The first complete range and the true cost of entering through the spread.
    What it decides
    Nothing yet — observation is the only permitted action.
    What it refuses
    Opening whipsaws, impulse and the urge to be first.
    Why it protects
    Waiting preserves tomorrow's options instead of spending risk on the market's least reliable minutes.
  3. PHASE 0309:45 New York (--:-- your time)

    1% VaR execution & partial exits

    At most one trade, only if the setup qualifies. Stop loss sized to the cent (1% max), automatic 50% take-profit at TP1 (+2.5%) and the stop moved to break-even at once.

    Read the logic of this phase
    What it observes
    Candidate quality, entry cost and the distance to the safety exit.
    What it decides
    One qualified position, or no position at all.
    What it refuses
    Unmeasurable downside, expensive entry and a second attempt.
    Why it protects
    The worst case is priced before the first share is bought, so today's opportunity cannot claim unlimited room from tomorrow.
  4. PHASE 0413:30 New York (--:-- your time)

    Hard cutoff · 100% cash flat

    Whatever is still open is liquidated by the server before the close. No overnight gap, no decay in leveraged products.

    Read the logic of this phase
    What it observes
    The clock and every position still open near the cutoff.
    What it decides
    Return the account to cash regardless of hope or opinion.
    What it refuses
    Overnight headlines, gaps and leveraged-product decay.
    Why it protects
    Tomorrow begins with options, not yesterday's unresolved risk. Capital can seek growth only after its continuity is protected.

When the market changes

How it notices a shift — and what it must prove first.

The engine is allowed to adapt, but only against days it has never seen.

  1. 01

    It reads today's context

    Volatility, macro calendar and rate pressure classify the session before any candidate is scored.

    How exactly

    A macro day moves 30 of the 100 points onto release deviation and the US10Y shift. A technical day moves them onto relative volume, the 15-minute opening range and the leading-proxy tape.

  2. 02

    It compares like with like

    Today is measured against past sessions in the same regime group, not against the whole archive.

    How exactly

    Sessions are clustered by volatility regime and behaviour. A cluster needs at least 30 intact days before it is allowed to influence anything.

  3. 03

    It has to prove itself blind

    A proposed change is fitted on part of the history and then tested on days it has never seen.

    How exactly

    80% of the sample is used to fit, 20% is held back untouched. A minimum edge and a walk-forward check must both hold, and any move is clamped to ±5%. Risk caps, the session flatten and the macro locks can never be touched.

See how the engine learnsExamine the results