Capitalise.ai

Capitalise.ai is a no-code trading automation platform that leverages natural language processing (NLP) to convert plain English trading rules into executable, backtestable strategies. Developed by Capitalise.ai (Tel Aviv, Israel) and launched commercially in 2019, it was among the first platforms to combine natural-language strategy building with automated broker execution. The platform targets retail traders, quantitative analysts, and family offices seeking to automate ideas without writing Python or MQL4/5 code.

Architecture & Technical Engine

The platform runs on a proprietary multi-model AI engine that combines Natural Language Understanding (NLU) models to parse English sentences into logical conditions, and a rule engine that maps these conditions to technical indicators, price levels, and time-based events. The engine operates on high-resolution historical data (tick-level and 1-minute) for backtesting, then routes orders to supported brokers via intermediary broker APIs. No public developer API is available, which limits external integration and makes the platform a closed ecosystem (walled garden).

Core Features & Performance

Users can write rules such as "Buy 100 shares of AAPL if the 50-day moving average crosses above the 200-day moving average and volume is above average," and the platform instantly converts them into a testable strategy. Features include historical backtesting, paper trading simulation, automated execution via brokers, real-time alerts, and support for technical indicators and scheduled rules. Performance is solid for simple-to-moderate strategies but degrades for complex multi-condition strategies or those requiring alternative data.

Integrations & Limitations

The platform integrates with brokers such as Interactive Brokers, TD Ameritrade, Alpaca, Tradier, and others, with a primary focus on US markets. Technical limitations include: English-only input, no public API, limited market coverage (mainly US equities), and incomplete support for crypto or forex in some plans. Output quality depends heavily on user phrasing accuracy, creating a non-traditional learning curve.