Eight gates, in order.

The process at a glance

Members move ideas through these gates on the Pipeline board. The board itself does not block a skipped gate; it records the skip as a process violation that the annual review reads back.

What each gate requires

1. Gate 0 · Mandate. Written limits before any research: maximum drawdown, maximum position, maximum sector weight and a cap on ideas in progress. Defaults are 25% / 8% / 30% / 10 ideas; members set their own.
2. Gate 1 · Regime. A monthly map of rates, concentration, flows and geopolitics, condensed into three written directives that constrain everything downstream (for example, a cap on AI-linked exposure).
3. Gate 2 · Screen. A cross-sectional quantitative screen (below) plus a disqualifier scan, a source channel, a “why now” paragraph and a devil’s-advocate question.
4. Gate 3 · Deep dive. A dossier scored 1–5 on six dimensions — business quality, moat durability, financial integrity, management quality, growth visibility and bear-case survivability. It passes only when the average is at least 3.5 and no dimension scores 2 or below: risks hide in the weakest dimension, not the average.
5. Gate 4 · Quant validation. Factor fingerprint, correlation with existing holdings, behaviour in past drawdowns, and days-to-exit at 20% of average daily volume.
6. Gate 5 · Valuation. A reverse DCF first — what growth does the price already imply? — then bear, base and bull three-year scenarios with probabilities. It passes when the probability-weighted IRR meets the market hurdle (defaults: US, EMEA and Asia-Pacific 12%, India 15%, plus 3 points for emerging markets), the bear-case loss is at most 25%, and the upside/downside ratio is at least 2:1.
7. Gate 6 · Stress and sizing. Twelve scenarios — six historical episodes and six hypothetical stresses — marked per position, summed across the portfolio and kept inside the drawdown budget with 20% headroom. Three to five numeric kill-switches are written before purchase.
8. Gate 7 · Execution. Limit orders, participation caps and a record of fills and slippage. Aurion does not execute trades; members use their own brokers.
9. Gate 8 · Feedback. Every decision is journaled with a confidence level. Outcomes are recorded later and compared with the stated confidence, so over-confidence shows up in the data before it shows up in returns.

The quantitative screen

The screen ranks a defined universe of listed equities across the US, EMEA, India and Asia-Pacific using market data from Interactive Brokers. It is refreshed on request, never on an automatic schedule, and each run records the universe version it used.

• Composite score = 0.40 × z(12-1 month momentum) + 0.25 × z(trend versus the 200-day average) + 0.15 × z(3-month return) − 0.20 × z(volatility), with z-scores taken across the universe.
• Five mechanical gates: liquidity of at least US$50 million traded per day, annualised volatility of at most 55%, price above the 200-day average, a 12-month maximum drawdown no deeper than −45%, and complete data.
• Results read “Passes screen”, “Fails screen” or “Watch”. They describe a filter, not a recommendation: a name that passes the screen has only earned a closer look at Gate 3.

The journal and calibration

The journal is append-only: entries cannot be edited after the fact. Stage moves, kills, kill-switch triggers, valuation overrides and process violations are written automatically; members add their own decisions with a confidence percentage. When a decision resolves, its outcome is recorded as a new entry. The Analytics page then plots stated confidence against how often decisions proved right, alongside violation rates, kill reasons by idea source and conversion through the funnel.

What this is not

Aurion Compass is a research and analytics tool. It does not give investment advice, manage money or execute trades, and screen results are not buy or sell signals. All investment decisions and their consequences remain with the member. Read the risk disclosure.