The engineering behind every order
A look inside Croftvex: where the data comes from, how the models learn and the safeguards every instruction passes through before it reaches the market.
From raw data to a clear signal
The engine draws on four streams at once: market data and price feeds, news and sentiment, macroeconomic indicators and order-book depth. Each potential trade is scored for both strength and risk.
Results arrive as buy, sell or hold signals with a risk read-out attached, then pass to allocation rules that you control. Models are retrained on a rolling basis, and every update is tested against historical data before it goes live.
Three stages, fully traceable
Every recommendation the engine makes can be traced back through the same three stages, so you can see why it appeared.
1. Ingest
Prices, volumes, order books, on-chain activity and news are collected continuously, cleaned and time-stamped so every model sees the same consistent picture.
2. Score
An ensemble of models rates each setup for direction, conviction and expected volatility. Disagreement between models lowers the score rather than being averaged away.
3. Check
Before anything is sent, the order is tested against your exposure ceiling, position sizes and stop rules. Anything outside your settings is blocked, not queued.
Protection in layers
Infrastructure is encrypted end to end and protected by mandatory two-step verification. Access to production systems is restricted, logged and reviewed.
The platform is monitored around the clock and checked through regular independent penetration tests, and client cryptoassets are held predominantly in offline cold storage. None of this protects you against falls in market prices.
Put the engine to work for you
Open an account to explore AI-assisted signals and automation with limits you set yourself.
Create account