About

We don’t predict markets.We detect anomalies.

Indian retail traders lose roughly ₹36,000 Cr annually — not because they lack access to data, but because they lack the computational capability to process it at institutional speed. We named that gap Processing Asymmetry. FinSight closes it.

Founder

Divyanshu Kumar

B.Tech, Christ University  ·  Founder, FinSight India

Built the first version of FinSight after watching retail traders consistently arrive late to the same patterns institutional algorithms had already exploited. Won 1st place at IIT Roorkee Cognizance 2026 IDEAZ Economics for identifying Processing Asymmetry as a distinct category of market inequality — separate from the information asymmetry described by Akerlof and Stiglitz.

Advisor

Prof. Sanjeev Sharma

Professor of Economics, Christ University

Co-author of the IMSICON 2026 paper formalising Processing Asymmetry. Provides the theoretical and economics framing that ensures FinSight’s detection methodology is grounded in market microstructure literature, not just statistical novelty.

Research

Processing Asymmetry: A New Category of Market Inequality in Algorithmic Trading

  • Presented at IMSICON 2026, Christ University
  • Co-authors: Divyanshu Kumar & Prof. Sanjeev Sharma
  • Status: Peer-reviewed, conference proceedings

Abstract.

We argue that a third form of market inequality — distinct from the information asymmetry of Akerlof (1970) and the asymmetric-information frameworks of Stiglitz — has emerged with the rise of algorithmic trading. We term this Processing Asymmetry: the inequality between actors who possess computational capability to act on identical, freely-available data, and actors who do not. Drawing on NSE and BSE tick-level data from 2020–2025, we estimate that this asymmetry transfers approximately ₹36,000 Cr per annum from retail to institutional participants — irrespective of directional market movement.

Request the full paper

Method

Statistical detection, not black-box AI.

Z-Score Anomaly

Every minute, FinSight computes the Z-score of each NSE-listed stock’s volume and price action against its 30-day rolling mean. When Z > 3, the deviation is statistically more extreme than 99.7% of the past month’s observations.

CAS Architecture

Beyond simple Z-scores, FinSight is migrating to a Composite Anomaly System that scores signals on five dimensions: statistical significance, persistence, cross-instrument confirmation, news/event correlation, and historical post-detection drift. Each dimension contributes to a confidence weight; only confluent anomalies surface.

  • • Volume Z-score (rolling 30-day window)
  • • Price-action Z-score (intraday OHLC)
  • • Volatility expansion vs. 14-day ATR
  • • Sector confirmation flag
  • • Outcome tracking (T+1, T+5, T+20)

Validation

1st Place — IIT Roorkee Cognizance 2026

  • IDEAZ — Economics Track
  • Cognizance, IIT Roorkee’s annual technical festival
  • Validated by panel of engineers, economists & domain experts

The competition required a fully working prototype, an economic justification grounded in published literature, and a defence against panel critique. FinSight placed first against entries from across India.

Read the data.
Then read the market.

Open the dashboard