Auron Labs

Fintech Research Lab

Build-in-public R&D for trading systems. We publish ambitious bets, failed experiments, and the lessons that hardened our production stack.

Experiment Flow

Experiments are documented with outcomes and lessons learned.

Live Monitoring

Active tracks are monitored continuously during iteration.

Research Corpus

Postmortems and technical writeups are maintained for major experiments.

Project Ledger

Failed, Ambitious, And Shipping Work

Real experiments, clear outcomes, and practical insights from systems tested under market pressure.

In Progress

Topological signatures and selective state dynamics for production-grade financial forecasting.

Why it mattered: Traditional sequence models were too brittle under regime shifts, costs, and multi-symbol leakage.

What happened: The pipeline now shows statistically significant edge in current OOS diagnostics and is in drawdown and cost-robustness tuning.

Lesson: Edge discovery is only phase one; deployment quality depends on disciplined risk shaping and symbol-level filtering.

Data Pipeline

Multi-broker normalization, calendar alignment, and event tagging for reproducible research at scale.

Infrastructure

Real-time health checks across strategy services, with high-fidelity logs and rollback-safe release stages.

Lab Timeline

Mar 2026

Postmortem published after repeated false positives in high-volatility windows.

Feb 2026

Introduced deterministic transforms to align replay and live diagnostics.

Jan 2026

Started comparing broker profiles by slippage distribution and reject rates.

Open Repositories

We publish the rough edges too: dead ends, rewrites, and bold hypotheses that did not survive contact with market reality.