How Treeova's Arch-AGI conviction engine works in plain language: 7 structured passes, regime-aware calibration, and what it intentionally is not.

    Back to Blog

    Arch-AGI Conviction Analysis, Explained — A Plain-Language Companion

    Treeova Research — Methodology TeamApril 18, 20264 min read
    Share:

    Companion post. Plain-language summary of the Arch-AGI Conviction Methodology whitepaper. For the full methodology, limitations, and what's intentionally withheld, read the whitepaper at /whitepapers/arch-agi-conviction-methodology.

    What is Arch-AGI?

    Arch-AGI is Treeova's deep-intelligence engine for evaluating an open options position. Instead of producing a single number that says "buy" or "sell", it runs a structured multi-pass review that asks the questions a disciplined human trader would: what's the edge, what could go wrong, what does the regime tell us, and how have we been wrong before?

    The output is a conviction score plus a written narrative — not a trade signal in isolation, and never a recommendation.

    The seven passes (high level)

    1. Edge analysis — what's the thesis, and what would invalidate it?
    2. Scenario projection — best, base, and worst case outcomes.
    3. Risk/Reward framing — breakeven, optimal exit, and stop-loss zones.
    4. Regime context — does this position fit the current market regime?
    5. Macro overlay — relevant tailwinds and headwinds.
    6. RL calibration — how have similar conviction levels played out historically, segmented by regime?
    7. Adversarial stress — a deliberate "challenge the thesis" pass that adjusts conviction down when the stress delta is large.

    What this is not

    • Not a recommendation. Reports describe a position; they do not tell you to open, close, or size a trade.
    • Not a backtest. Conviction scores reflect current context, not historical replay.
    • Not infallible. The adversarial pass exists precisely because the earlier passes can be wrong.

    Limitations

    • Past calibration does not guarantee future calibration.
    • Reports depend on the quality of the upstream intelligence feed.
    • The system is intentionally conservative when data is sparse for a given regime.

    Read the methodology

    Arch-AGI: Seven-Pass Conviction Methodology →


    Past performance does not guarantee future results. See /legal/risk-disclosure.