Turn a thesis into a document you can inspect.

A Strategy document holds the nodes, edges, market context and execution settings that make a rule inspectable. The graph is not just a visual sketch: it is the executable identity that later surfaces must reference.

Keep the document small enough to explain. A strategy that cannot be described from its sources, conditions, sizing and exits is difficult to review no matter how polished the graph looks.

• SourcesMarket and indicator inputs that provide the values the graph can use.
• OperationsComparisons, logic and transformations that make the thesis computable.
• Execution settingsTimeframe, sizing, entry, exit and risk boundaries that define how the rule is evaluated.

A backtest tells you how the frozen logic behaved against the selected historical assumptions. A chart preview tells you whether the graph reads as intended in the current view. A Paper automation tells you what the runtime actually observes and records.

Compare those layers deliberately. Look for differences in timeframe, data availability, fees, funding, fills, slippage, position sizing and execution timing before calling an edge real.

The verdict on a durable run — candidate, hold, reject — is an annotation you wrote, not a computed result. The run page shows the verdict it currently carries and, beside it, every change that verdict has been through: what it moved from, what it moved to, and the note that explained the move.

Runs that were annotated before this history existed carry a single row labelled as such. The earlier judgements are genuinely unavailable, and the page says so rather than reconstructing them from timestamps.