IROHA for prop and quant desks
For a desk, the useful question is whether a move was the market, the sector or something specific to the company, and whether that kind of cause tends to persist. IROHA labels moves by cause with evidence scores and exposes them as features your own models can consume.
Cause-labelled moves as model features
Each large move comes with its market, sector and company-specific components and the event type behind the residual. Use them as features or as labels for your own research.
Filtering index noise from company shocks
Separating the market and sector part from the residual lets you study company events without the index move drowning the signal.
API access and latency
Pilot customers get API access. Output is computed after each session; the API is built for research workflows, not order routing.
Index insights: Bank Nifty today
The API covers the 14 Bank Nifty constituents and the index today. Wider universes run in pilots.
What the output looks like
Take RAYMOND, which rose +52.8% from 21 Aug – 7 Oct 2026. IROHA puts −0.5 points on the market and −1.9 on textile peers, names the dated cause, and shows which drivers pass every causal test: policy news and company news. Across all 52 big RAYMOND moves since 2015, the model leaned the right way before 41. The full study.
See how IROHA works and a worked example on RAYMOND.
Questions
What format is the output?
JSON over the API, keyed by stock and date, with the market, sector and company-specific components and the evidence tests for each cause.