​Anomaly detection is generally used to automatically detect out-of-distribution behaviours in asset prices, which may be the result of undisclosed new information, ripple effects from correlated assets, or structural breaks in an asset’s behaviour versus the broader market.

​

Although these anomalies do not represent a directional indication regarding future price moves, they are predictive of impending volatility events and as such are powerful risk management tools.

​

As part of our embedded collaboration with Axyon AI, a leading European FinTech with expertise in Deep Learning AI for asset management and trading firms, DeFinity has been providing its customers and tokenholders access to premium crypto market anomaly detection scores since September 2021.

​

The technology has proven extremely valid to monitor crypto market data and generate early warnings regarding:

​

(i) unexpected price behaviours,
ii) anomalous liquidity changes, and
(iii) inconsistent transactions that could be a sign of broader volatility events on price.

​

On an hourly basis, approx. 5 minutes past each hour (UTC), a new anomaly score (0-100%) is generated by deep learning AI models for each of the 12 supported crypto assets, (AAVE, ADA, BNB, BTC, DOT, ETH, LTC, MATIC, SOL, UNI, XMR, XRP). The deep learning AI models look at a combination of market data, context data and on-chain data to detect irregular patterns.