Quandora Workflow

Five public Quandora skills, their evidence boundaries, and the explicit confirmations between them.

Quandora separates creation from analysis and historical testing from simulation. The public workflow is implemented through five skills:

Factor Mining -> Factor Analysis -> Strategy Building
      ^                |                   |
      +-- user-approved factor experiment-+

Strategy Building -> Strategy Analysis -> Paper Trading
        ^                    |
        +-- user-approved strategy experiment

Analysis is read-only. A proposed factor experiment returns to Factor Mining; a composition or configuration experiment returns to Strategy Building. Paper Trading begins only after a separate explicit confirmation.

All five skills are available to public users. Paper Trading is simulated. Live trading is not part of the public product. See Product Availability.

1. Factor Mining

Factor Mining starts from a public Research Task or a custom idea. The agent reads the server contract, creates and validates a contract-compliant plugin.py, checks for materially similar work, submits the exact validated source, and follows the returned run.

Output: one factor result and, when requested and available, one verified Result Bundle ZIP.

2. Factor Analysis

Factor Analysis selects one exact run and reads owner-scoped, server-persisted IS evidence. It diagnoses the Factor Card, Health, rating fields, chart data, and inert source when needed.

Output: observations, inferences, alternatives, risks, and controlled experiments. It does not submit or modify anything.

3. Strategy Building

Strategy Building lists eligible Official, Mine, and Shared factors. It composes a public cross-sectional Strategy, shows the effective configuration, obtains confirmation, submits the backtest, follows the same run, and can explicitly export its verified Result Bundle.

Output: one canonical Strategy run with its exact factors and parameters.

4. Strategy Analysis

Strategy Analysis selects one exact run, reads its canonical configuration and retained artifacts, and uses bounded numerical chart evidence for diagnosis.

Output: performance, risk, turnover, exposure, attribution, factor-combination, and Paper-readiness analysis. It does not submit a new Strategy or Paper run.

5. Paper Trading

Paper Trading selects an eligible completed Strategy source and obtains a separate confirmation before starting simulated execution. It can inspect the current portfolio, closed positions, fills, funding, equity, and bounded code, and it can stop the run after explicit confirmation.

Output: simulated execution evidence. A stopped run is terminal and cannot be resumed.

Explicit Handoffs

From	To	Required user decision
Factor Analysis	Factor Mining	Choose a proposed factor experiment and confirm a new run
Factor or Strategy evidence	Strategy Building	Choose eligible factors and confirm the exact configuration
Strategy Analysis	Strategy Building	Choose a proposed composition/configuration experiment
Eligible Strategy source	Paper Trading	Confirm the exact source and simulation settings
Paper evidence	Any research skill	Choose the next action; no automatic restart or repair

Local Outputs

Analysis does not depend on local files. In writable hosts, explicit completed exports use:

Quandora result/factor/<factor_slug>.zip
Quandora result/strategy/<strategy_slug>.zip

Each verified ZIP remains intact and is not automatically extracted or rebuilt. Its runtime manifest is the authority for included, pending, and omitted items.

Safety Boundary

Backtests and simulated Paper runs are evidence under defined conditions. They are not future-return guarantees, financial advice, or permission for real-money execution. Missing evidence remains missing, analysis remains read-only, and mutations require explicit confirmation.