Research process

Evidence moves through
fixed gates.

The process expands coverage without merging research, review, and portfolio authority.

From signal to decision

Six stages preserve the reasoning.

Each stage produces a durable output for the next stage.

01

Build the opportunity set

Candidates enter through systematic scans, macro narratives, or company events. Each candidate keeps its original source and timestamp.

The process screens a broad universe. It does not rely on recalled names or unsupported familiarity.

02

Assemble the evidence

The research packet joins price behavior, company facts, valuation, catalysts, and regime context.

Each material claim carries a source. Gaps remain visible instead of receiving an invented answer.

03

Write a falsifiable memo

The memo states the thesis, the opposing case, the expected path, and the evidence that breaks the view.

A complete case separates observed facts from analyst judgment. It also defines the review horizon.

04

Apply independent challenge

A separate reviewer checks source quality, internal logic, valuation support, and the kill condition.

The reviewer can approve, return, or reject the case. The reviewer cannot change the portfolio directly.

05

Rank within the portfolio

Approved cases compete against holdings and cash. The portfolio process weighs conviction, timing, and related exposure.

A good company does not guarantee a position. The case must improve the portfolio at its current price.

06

Monitor and learn

Scheduled reviews compare new evidence with the original case. Each change keeps its reason and effective date.

Post-trade reviews separate process quality from market outcome. Useful lessons become testable hypotheses.

Evidence protocol

The record must support the decision.

The system treats source discipline as part of the investment method.

Protocol

Point in time

Use information available at the decision date. Preserve publication and event dates.

Protocol

Source quality

Prefer primary evidence. State when a secondary source or estimate remains necessary.

Protocol

Fact and judgment

Keep observed data separate from interpretation. Label uncertainty in plain terms.

Protocol

Reproducibility

Store the inputs, rules, and decision record needed to inspect the result later.

Role separation

No role grades its own work.

Research builds the case. Review tests it. Portfolio judgment allocates. Execution applies the approved state.