Talk to someone
who understands markets.
It is not a chatbot that talks about trading. It is an agent with over 100 MCP tools that fetches real data, calculates indicators, trains models and runs backtests — all through natural language.
Verida — the model executes real market tools, not just answers
How it works in practice
Fetch the last 500 1h candles of BTCUSDT
The AI dispatches the buscar_binance tool, persists as raw series, and responds with the URI + metadata.
Calculate 14-period RSI and MACD
The AI chains rsi() and macd() over the series, persists as derived, and you get the URI + latest value.
Train a classification model with RSI, ATR and volume
The AI builds the ML DAG with feature engineering, temporal split, training and validation — all via tools.
Run a backtest of the moving average crossover strategy
The AI loads the Rhai strategy, injects indicators, executes the backtest and returns Sharpe, drawdown and PnL.
Supported models
You bring the key — free or paid. Verida does the rest.
Anthropic Claude
Anthropic API key
Recommended for best understanding of the 100+ MCP tools.
OpenAI GPT
OpenAI API key
Works perfectly with MCP tools.
NVIDIA NIM (free)
NVIDIA API key
Cutting-edge models via free NIM API. Generous pool limits — zero cost to start. Verida default provider today.
Ollama (local)
Local model via Ollama
Llama 3, Qwen, Mistral running locally. Zero API cost, full privacy.
Google Gemini
Google AI API key
Works with MCP tools.
Other providers
Any compatible endpoint
OpenRouter, Together AI, Groq, DeepSeek — if it speaks OpenAI-compatible, it works. You choose the provider — free or paid, it is your decision (or limitation).
MCP: the standard that structures AI
Model Context Protocol is what allows the LLM to not just "know about" markets, but execute market tools. Verida exposes over 100 MCP tools — fetch data, calculate indicators, train models, run backtests, query microstructure. The model decides which tool to call; you decide what to ask.
Without MCP: the LLM hallucinates data. With MCP: the LLM executes. The difference is everything.