Reasoning Model
Leverage time-based reasoning for financial time series modelling.
Building on our base models, these models use time-based reasoning to dynamically adjust learning rates based on time-varying market relationships. This enables adaptation to changing market conditions to predict market trends—giving more weight to recent information during volatile periods while maintaining stable long-term perspectives during normal conditions.
This visualisation shows the model's time-based reasoning capabilities for BTCUSDT, showing trend identification on both 1-hour and 4-hour timeframes from 2020 to 2025 with no prior market specific training.
This visualisation shows the model's time-based reasoning capabilities for the VIX Index, showing trend identification on both 1-minute and 5 minute timeframes for 2024 with no prior market specific training.
While these systems utilise proprietary reasoning algorithms, they have not yet been fine-tuned with real-world market data. Users can significantly enhance predictive accuracy through custom multi-time-horizon fine-tuning or by incorporating their own proprietary datasets.