Discover tools for AI investment research

Hosted service using a simulated fruit-fly connectome to decide trades for per-NFT wallets; activity is logged publicly

The Hugging Face model card identifies AdaptLLM/finance-LLM-13B as the base model; the repository provides quantized weights for local inference toolchains.

Uses multi-timeframe indicators for portfolio decisions while application logic enforces sizing and TP/SL rules

The model card says it maps sentences and paragraphs into a dense vector space for clustering or semantic search in RAG applications.

Its training environment was designed for tick data and also supports OHLC data; the README lists real-time trading as a future direction.

The model card says it was fine-tuned on stock-market news headlines and predicts positive, neutral, or negative sentiment.

The Hugging Face model card classifies it as a time-series forecasting model and provides PyTorch model artifacts.

The model card says it is fine-tuned from FinBERT to classify tweets into 20 financial topics.

Optional Jev provider for structured judgments in multi-agent US-equity research and backtests

AI trading agent spanning research, execution, and position management across several asset classes

Hyperliquid trading skills paired with Hyperfeed trader data

It presents model trading activity and profit-and-loss results using real market data, following the NOF1.ai Alpha Arena format.

Based on TradingAgents, the project implements and reproduces the framework for the A-share setting, with a research and teaching focus.

It arranges analyst, researcher, and risk roles into a collaborative research flow for examining market information, forming investment hypotheses, and seeing how different roles reason about a decision. It helps explore multi-agent workflows for financial analysis.

Who it suitsDevelopers exploring multi-agent financial research who can configure model and data services.

Composable environments, action schemes, reward functions, and data feeds can be combined to build, train, and evaluate algorithmic trading agents.

Experimental two-phase trading template that defaults to dry-run, with live execution opt-in

It brings together watchlists and holdings, daily reviews, industry news and research, stock deep-dives, backtests, and research reports.

LLM trading terminal with a Go runtime that enforces order risk limits

Use agents to organize research and development tasks; its quantitative scenarios include iterative exploration of factors or models. Researchers can inspect the project scenarios and demo to understand the workflow, then assess fit for their own experiments with task-specific configuration.

Who it suitsResearchers exploring agent-assisted workflows who can configure the environment and model services.

Optional pre-trade Jev decision gate before the platform execution flow