deciqAI RIA — Position Sizing & Concentration Discipline
Industry front door for expected-value-and-kelly. Adds domain triggers, example, packs only. Parent Process unchanged. Not investment advice. Educational framework; apply within IPS and suitability.
Activate when: sizing a single-name or thematic position; a client with concentrated stock; setting max-position/rebalancing bands; "how much of the portfolio should this be?" Do NOT activate when: broad index allocation already governed by the IPS.
Why this variant
The parent expected-value-and-kelly sizes repeated positive-edge bets to maximize long-run growth without ruin. For an RIA, the discipline is inverted toward survival: fractional-Kelly thinking caps concentration so no single position can impair the client's plan.
Domain inputs → parent's Process
- Establish the edge is real and repeatable before sizing anything.
- Use fractional Kelly (e.g. ¼–½) — full Kelly is too volatile for client capital.
- Bound by IPS max-position and drawdown tolerance; the smaller of Kelly-implied and IPS cap wins.
- Treat concentrated legacy stock as an over-sized position to unwind on a tax-aware schedule.
Worked example
Client insists on a 25% single-stock position. → Even a generous edge under fractional Kelly rarely justifies 25% of a household's net worth; ruin risk dominates. Frame the trade-off, cap per IPS, and set a diversification glide-path (tax-aware). Documented as a suitability decision.
Packs
- Solo: max-position + rebalancing-band rules in the IPS.
- Advisory: concentrated-stock unwind playbook (tax-aware).
Red flags
- Sizing to full Kelly / conviction with client capital.
- Concentration exceeding IPS caps "because it's been working."
- No ruin/drawdown check.
Verification
- Edge validated before sizing
- Fractional Kelly, bounded by IPS cap (smaller wins)
- Drawdown/ruin tolerance respected
- Concentration decisions documented for suitability
Part of deciqAI Knowledge Skills. Core method: expected-value-and-kelly.
Part of deciqAI Knowledge Skills — 237 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/s/ria-position-sizing-kelly · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/ria-position-sizing-kelly.json