I built an AI that tells you exactly what stocks to buy, how much, and when to sell

Thanks! Trust and clarity are definitely top priorities — especially in finance where overpromising is rampant.

On "signals" transparency:

• We're explicit that these are ML-ranked probability scores, not predictions. The models output relative rankings (which stocks look stronger vs weaker based on fundamental + technical features), not price targets.
• We surface validation metrics publicly: Spearman IC, hit rates, and long-short spreads. Users can see when models are performing within healthy thresholds vs degraded.
• Health scores (A+ to F) for personal portfolio analysis help users understand relative positioning without implying guaranteed outcomes.

Onboarding approach:

• Tiered strategy access (Starter → Professional → Enterprise) lets users start simple before accessing more complex strategies
• Position changes are surgical (1-5/day) with clear audit trails, so users see exactly what changed and why
• We show entry dates and period returns vs SPY benchmark — grounding everything in measurable, comparable outcomes rather than abstract "AI picked this"

Feedback loops we're watching:

• Subscription retention by tier (are users finding value?)
• Portfolio upload → analysis completion rates
• Which AI insight tabs users actually click (Market Regime, Performance, Sectors)
• Email notification engagement (position change alerts vs weekly digests)

The honest answer is we're still early on formalizing this — the product grew from ML infrastructure outward. Your point about early metrics shaping the roadmap resonates. What patterns have you seen work well for building that trust layer?