Assets Risk Analyser

First See, First AIM, First Shot

Comment

Project Overview

Project Name: Asset Risk Analyser

Elevator Pitch / Tagline: Forecasting financial bubble bursts and macro risks using AI vector memory matching and real-time asset analytics.

Project Story

Inspiration

Why do market crashes like the 2008 Lehman Brothers crisis, the 2021 crypto bubble, oil shocks, or the Dot-Com crash keep repeating? It is rarely because investors are careless—it is because of a critical lack of accessible, actionable information. Most individuals cannot aggregate, synthesize, and analyze complex market data from disjointed sources in real time. Asset Risk Analyser bridges this gap by democratizing institutional-level macroeconomic analysis.

What It Does

Asset Risk Analyser continuously ingests financial data for monitored assets and processes it using AI to deliver clear, graphical insights:

Historical Pattern Matching: Uses vector memory to map current market parameters against historical crash precedents (e.g., matching leverage and valuation metrics against the 2021 SPAC peak or 1999 Dot-Com bubble).

Early Warning Alerts: Detects speculative spikes and overvaluation indicators to alert users to impending bubbles before they burst.

Macro Scenario Simulations: Calculates probabilistic risk projections showing how global economic events and systemic liquidity shocks could impact individual portfolios.

How We Built It

The architecture relies on a decoupled full-stack design split across two dedicated repositories:

Frontend: Built with React, Vite, and TypeScript (crafted via Lovable) to deliver a dark-mode dashboard featuring risk meters, trend graphs, and real-time alert feeds.

Backend: Powered by Python (FastAPI), orchestrated in GitHub Codespaces with GitHub Copilot for rapid debugging.

Database & Vector Storage: CockroachDB powering vector search via 768-dimensional Gemini embeddings to match current asset profiles with historical crash regimes.

Data & AI Engines: Real-time asset fundamentals retrieved via yfinance, powered by high-speed Groq and Google Gemini inference engines.

Challenges We Ran Into

The most demanding obstacle was end-to-end integration between the microservice backend and the React frontend—specifically resolving strict payload schema mismatches. Gemini embeddings and model outputs returned specialized array representations (such as 768-dimensional float vectors) that mismatched the expected types on the client side. Debugging these serialization pipelines across independent repositories required an entire day of deep API response reshaping and transformation logic.

Accomplishments That We're Proud Of

As someone stepping into end-to-end full-stack integration side-by-side for the first time, managing live deployment synchronization across multiple active browser tabs, database console portals, and split terminal sessions was an empowering milestone. Seeing a complex multi-layered pipeline function smoothly to help everyday users protect their financial capital made every hour of effort worthwhile.

What We Learned

Cloud & Vector Data Pipelines: How to generate embeddings locally, persist them to cloud-native databases like CockroachDB, and query vector similarity in real time.

Microservices Integration: The nuances of type-safe payload mapping between decoupled Python FastAPI backends and TypeScript frontends.

Systemic Resiliency: How small formatting oversights or API shape drift can halt an entire dashboard pipeline, highlighting the importance of strict interface boundaries.

What's Next for Asset Risk Analyser

Live Market Manipulation Simulation: Interactive sliders allowing users to stress-test their portfolios against custom interest rate hikes, liquidity squeezes, or geopolitical shocks.

Custom AI Dashboards: Modular, user-configurable AI dashboards where multiple asset classes and macros can be simulated simultaneously on a single unified canvas.

Built With

Languages & Core: python, typescript, javascript

AI & Machine Learning: groq, google-gemini, vector-embeddings

Backend & Database: fastapi, cockroachdb, yfinance

Frontend & Tooling: react, vite, lovable, github, github-codespaces, github-copilot

Built With

Share this project:

Updates

Submission history