AI Research guide¶

AI Research organizes market data, instrument context, and model analysis into a reviewable research process. It can speed up research, but it does not replace data checks, risk judgment, or investment decisions.

Start a research task¶

1. Select the correct market and instrument; avoid a similarly named product on another venue.
2. Check quote time, currency, venue, and data frequency.
3. Choose a goal such as trend, fundamentals, risk, or cross-instrument comparison.
4. Run the analysis and wait for data and model stages to finish.
5. Save useful findings or continue into indicator and strategy development.

Decide whether a result is usable¶

• Data time: check quote, filing, and news timestamps; do not present old data as current.
• Evidence: model conclusions should agree with charts, indicators, and structured data.
• Market identity: the researched instrument must match the intended execution product. Exchange equities, tokenized equities, and traditional broker securities are different identities.
• Missing data: shorten the range, change provider, or stop when required history or fundamentals are absent. Empty data is not a no-signal result.
• Model variation: models may disagree. A person should review important conclusions.

Continue from research to a strategy¶

Research can define a hypothesis, screen a universe, or produce a code draft. In the strategy editor you still need to:

• Define entry, exit, sizing, and risk rules.
• Use canonical market identifiers and data frequencies.
• Pass Strategy API V2 validation.
• Backtest on an independent range and inspect execution assumptions.

Continue with the strategy workflow. See the indicator development guide for indicator contracts.

Result boundaries¶

AI output may contain errors, omissions, or unsupported inferences. Remove account details, secrets, and personal data before sharing a report. Before live use, verify source data, product identity, costs, liquidity, and account risk.