edgedepth EARLY ACCESS
Open terminal Research Workbench
Menu
Product Product overview Live terminal Tick-by-tick replay Liquidation heatmap What EdgeDepth is Compare Roadmap Research Research hub Research Workbench Investigate a move Published reports Daily brief Event archive Blog Learn Guides Reading library Worked examples Guided lessons FAQ Developers API & AI agents API documentation MCP setup Open source Pricing Open terminal Research Workbench

Research

Does your trading idea hold up?

Check what happened after similar setups. Understand the failures, then decide whether the idea deserves more work.

Open the WorkbenchRead a completed study →

  1. Define the test

    State the exchange and selected markets, historical dates, exact conditions and how long after the setup to measure.

  2. Compare what followed

    Count the completed outcomes and missing data. Compare with the same markets and dates without requiring the setup.

  3. Choose the next check

    Look for failures. Test a separate period, refine the question or stop pursuing the idea.

The comparison includes all eligible minutes, including setup matches. It is a reference for your investigation, not a prediction or a trading return.
An explicit EdgeDepth research definition with conditions and scope
AN INSPECTABLE DEFINITION The exact conditions, markets and dates define what the result can say.

Start with an observation

Describe a setup or investigate a historical move. The Workbench turns a supported question into conditions you can review and edit. Combine conditions with AND or arrange an ordered sequence, then choose the markets and dates.

Available readings include candle structure, price movement, funding, open interest, liquidations and recorded order-flow measurements. The reading library explains their definitions and limitations. Not every reading exists throughout every market history.

Check all the matches, not just the winners

Run the definition you approved. See how often the setup appeared and what happened next. Each percentage shows its count. Missing outcomes are listed separately; the examples you browse are only part of the result.

Where available, compare with what price did across the same markets and dates without requiring the setup. This reference includes matching minutes; it is not a separate control group. Nearby matches can overlap, so a large count does not necessarily mean many independent events.

Look for reasons the idea could fail

Inspect favourable and adverse outcomes, concentration in individual markets or periods, and cases that contradict the idea. Replay the recorded trades and book changes where the history and your plan allow.

Revise the definition and compare runs to understand what changed. Separately detected studies can have different anchors and populations. A change in their outcome rates does not establish that the added condition caused it.

Decide what to check next

A negative finding can help you stop pursuing an idea. A thin sample can justify a different period or more observations. An interesting result can justify a separate-period check with the setup held fixed. None is automatically a profitable strategy.

Save a study when you want to return to the definition and evidence. Monitoring is available through the web workflow for supported definitions and requires confirmation. It applies those conditions to future closed market minutes; an alert means the condition occurred again.

How does research help a trading decision?

It can help you reject a weak belief before risking money, identify the losses hidden by a winning example, or write clearer rules to investigate further. An interesting pattern still needs checks for fees, execution, risk and a separate period. A historical price move is not the profit you could have captured.

Use the interface that fits

The browser, REST API and MCP use the same research engine. Machine-readable results retain exact query and revision metadata. MCP can prepare and run approved research; saving and alert management remain web actions.

Check coverage before interpreting the result

Research currently covers Binance USDT-M. Studies can reach the full verified record where their required measurements exist. Historical research and tick replay have different coverage and access. Inspect available data and plan limits.

Reading an existing public report uses no research allowance. New or materially changed studies can consume credits according to scope. For the full workflow, read how research works.