Loopera-ai/loopera

What it solves

Loopera is an AI agent designed for fundamental quantitative research. It solves the problem of transforming raw financial data and research materials (like PDFs and prompts) into validated, explainable investment factors. Instead of randomly combining data fields to find statistical correlations, it focuses on the economic logic behind financial anomalies to ensure that research results are grounded in business reality rather than data fitting.

How it works

Loopera operates as a continuous research loop consisting of several stages:

1. Input & Understanding: It accepts research topics, text prompts, or PDF reports, extracting claims, formulas, and economic mechanisms.
2. Hypothesis Generation: It identifies financial anomalies (e.g., profit not converting to cash) and proposes competing economic explanations to explain why these phenomena might impact future performance.
3. Factor Construction: It converts these hypotheses into "Research Contracts"—structured, computable objects that define the expected direction and economic meaning of a factor.
4. Evidence-Gated Validation: Candidates must pass through multiple "Evidence Gates" (checks for data availability, timing/look-ahead bias, accounting consistency, and incremental value) before being accepted.
5. Research Memory: Both successful and failed research paths are stored in a memory system to prevent redundant effort and identify research gaps.

Who it’s for

• Fundamental Quantitative Teams: Those looking to standardize and expand their factor research process.
• Investment Research & FinTech Institutions: Organizations wanting to connect internal reports and data capabilities to a standardized validation pipeline.
• Research Managers: Those who need to track the decision-making logic, evidence, and failure reasons behind research results rather than just seeing backtest curves.

Highlights

• Hypothesis-Driven: Prioritizes economic explanation over raw computation to avoid over-fitting.
• Evidence-Gated: Uses a multi-layer validation process to ensure factors are robust and logically consistent.
• Memory-Augmented: Maintains a research lineage that records why certain paths failed, turning failures into team knowledge.
• Multi-Modal Input: Can ingest PDFs to extract research claims and formulas while maintaining strict source fidelity.