Research & evidence

Give every idea a reason to be tested.

Research starts a hypothesis. You decide what becomes an experiment. Studio prepares focused queries for your browsing-capable AI client, then checks the completeness and relevance of the source records you supply. It does not retrieve or authenticate publications.

Match research to your strategy.

Describe your idea or paste a code excerpt. Open Research sources to choose where your AI should look or add specific paper links. Prepare your brief, then copy it into ChatGPT or Claude. You can attach your complete code in that conversation.

Starting point

Brief preparation in your browser. This form does not submit your text or call AI. A privacy-settings reload temporarily keeps your current brief in this tab.

Open the full strategy prompt builder

See why a paper fits—and where it does not.

Your browsing-capable AI finds and reviews sources. Studio compares the supplied study mechanism, market and timeframe with your strategy, then returns ranked matches, evidence gaps and the next test. Ideas from another market or horizon are marked as conceptual transfers.

The transparent score separates strategy fit from supplied evidence details. Missing information earns no points. Matching uses metadata and keywords, so review the original paper before approving a change. Studio does not independently authenticate sources, and a high score does not predict trading performance.

Sources to explore

Working papersSSRN

Finance hypotheses and early research.

Quantitative financearXiv q-fin

Microstructure, statistics, and machine learning.

Economics researchNBER

Macro regimes, risk premia, and policy effects.

Academic evidencePeer-reviewed journals

Methods, limitations, and replication context.

Strategy researchQuantpedia

Anomalies, implementation notes, and original papers.

Practitioner replicationAlpha Architect

Stress tests, critiques, and implementation risks.

Institutional researchCFA Institute Research

Portfolio construction and practitioner evidence.

From evidence to experiment

A research finding becomes a separate candidate with an economic rationale, assumptions, failure conditions, and a validation plan. You approve the experiment before it becomes part of your strategy.

Evidence suggests what to test. Your approval comes before implementation; fresh backtest results and robustness checks come before a final selection.

Keep the validation loop complete

Generate candidate code, audit its exact rules, compile and backtest in your runtime, and bring fresh results to Studio. Evaluate costs, out-of-sample evidence, and Monte Carlo before revising or selecting a candidate.

Read the full validation workflow ↗