The project aims to design and evaluate a system capable of identifying emergent behavioral signals from financial market data—specifically through unsupervised exploration of cross-asset patterns—without reliance on hard-coded assumptions or predefined factors.
Our technology simplifies 16 million securities—spanning all financial products worldwide—into a clear, standardized, and easily understandable commodity. This ensures absolute clarity and transparency, free from bias. Our innovation delivers data that is standardized, regulated, pure, and intelligent, making it an optimal source of high-quality financial information for AI.
Following qualifications required:
- Data handling (EOD, normalized values) – EOD data handling, time series familiarity
- Abnormality Detection (unsupervised ML) – Isolation Forest, One-Class SVM, Autoencoders, DBSCAN
- Early Signal Detection (lookahead modelling) – supervised ML, lookahead design, transformer models
- Feature Engineering (multi-parametric) – parameter sweeps, rolling window features, drawdown, volatility
- Product characteristics – financial time series familiarity; minor gaps without domain guidance
- Clustering & Grouping – experience with clustering methods, correlation-based distances
- Signal Recognition/Definition Framework – covered in practice, for financials
- Signal Evaluation & Scoring – understanding; metrics like anomaly strength, impact likely covered
- Textual Description of Signals (LLM-based)
