บล็อก

In-depth review of Fincept Terminal v4 — a native desktop application built on C++20 and Qt6 with 37 AI agents, QuantLib, and 100+ data connectors for professional trading.

Review of Kronos — a foundation model for OHLCV candle forecasting. BSQ tokenizer, hierarchical decoder, two-stage sampling, Qlib training pipeline. How a model learns the 'language' of the exchange.

In-depth review of Jesse — an algo-trading framework for crypto markets. Minute-based simulation, strategies as state machines, Rust-accelerated indicators, optimization with overfitting protection, and the boundary between open-source core and live trading.

A deep dive into AI Hedge Fund by virattt — an open-source system where multiple LLM agents with different analysis styles build a portfolio through a risk filter. Architecture, agents, limitations, and lessons for real systems.

Complete guide to the AI4Finance Foundation ecosystem: FinGPT (LLM + LoRA for finance), FinRL (reinforcement learning for trading), FinRobot (multi-agent orchestration). Satellites, pipelines, and practical usage.

Overview of VectorBT — an innovative quantitative analysis library that changes the approach to backtesting thanks to the power of NumPy and Numba.

Beyond linear correlation — using copula models to capture tail dependence and joint risk in cryptocurrency portfolios for accurate VaR and CVaR estimation.