A research-only model that predicts next-period direction for INFY using a hybrid of traditional ML + attention with domain constraints, so the signal is explainable (top contributors are shown, not hidden in a black box).
Built on
Regulatory data obtained for research and PoC modeling only.
Current result: Rolling directional accuracy ~59% on the regulatory dataset
Next Steps: Increase coverage and stability by adding additional causal data streams; broaden regime tests.
Status: PoC only. Not for live trading. This dataset was used solely to validate causal modeling.
In Progress (Two more prototypes)
VIX Anomaly Detector
Detects volatility spikes / rare regimes; built to flag abnormal conditions early. (Research-only.)
R.A.V.A.N — Market Surveillance
Watches ~200 assets simultaneously for cross-asset anomalies and structural breaks. Outputs under IP at this stage. (Research-only.)
Why Investors Need to Watch
Attention is all the machine needed to move up from traditional algorithmic ML. Our Hybrid machines demonstrate causal awareness while being Explainable for the Regulatory bodies.