Categories
Self-Hosted AI Agents MCP Finance Backtesting Algorithmic Trading Quantitative Research LLM AppsStars
Forks
Watchers
Developer links
Vibe-Trading
Ask for a backtest of a 20/50 moving-average crossover on BTC-USDT for 2024, and Vibe-Trading plans the run, fetches the bars, writes the strategy, and returns a run card with drawdown, trades, and a buy-and-hold comparison. The web chat keeps every session, accepts uploaded filings and spreadsheets, and links each answer to a Run Detail page with the equity curve and the data source used. Eleven backtest engines cover A-shares, global equities, crypto, futures, forex, options portfolios, and cross-market composites, and the alpha zoo benchmarks factor libraries like GTJA191 across a chosen universe. Swarm teams (investment committee, quant desk, risk review) split a question among macro, quant, and catalyst workers and merge the result into one report. Paper trading tracks cash, fills, and NAV between sessions, the portfolio view proposes allocation changes, the options page draws multi-leg payoffs and Greeks, and scheduled playbooks repeat a brief on a timetable. The Shadow Account reads broker trade journals, extracts implicit rules, and replays your trades against a rule-based alternative. Model providers include OpenRouter, OpenAI, Anthropic, DeepSeek, and local Ollama, and any broker execution is opt-in within limits you set and can halt. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Benefits
- Backtests From Plain English
- Describe a strategy in a sentence and receive a runnable backtest with trades, drawdown, and a benchmark comparison, so testing an idea no longer starts with writing code.
- Evidence Behind Every Answer
- Each run card records the data source, dates, and caveats, and research artifacts are versioned with dated revisions, so a conclusion can be revisited and compared later.
- Coverage Across Many Markets
- Twenty-eight data sources with automatic fallback serve equities, crypto, futures, forex, and options, so one workspace can compare an A-share index with a Bitcoin strategy.
- Your Models, Your Server
- Choose OpenRouter, OpenAI, Anthropic, DeepSeek, or a local Ollama model, keep sessions and runs in persistent volumes, and reach the workspace from any browser.
Features
- Run Detail Reports
- Open a saved backtest to see the equity curve, trade list, validation warnings, and data source, then ask a follow-up with different parameters.
- Swarm Research Teams
- Preset teams for investment committee, quant desk, and risk review assign workers to macro, quant, and catalyst questions and combine their findings into one report.
- Alpha Zoo Benchmarks
- Compare factor libraries like GTJA191 over a chosen universe and period, and rank the top alphas before testing one on your own data.
- Paper Trading And Portfolio
- Simulated accounts persist cash, fills, and NAV between sessions, and the portfolio view combines holdings and proposes allocation changes for review.
- Options And Correlation Tools
- Build multi-leg option payoffs with Greeks, and explore asset correlations across market regimes from dedicated pages in the web app.
- Shadow Account Review
- Import a broker trade journal, extract the rules implied by your trades, and run counterfactual backtests that render an audit report.