How I Built a Real-Time Cricket Arbitrage Scanner (Bloomberg Terminal Style)

#cricket #javascript #machinelearning #webdev

The Idea

During IPL season, I noticed something interesting — different bookmakers often disagree on match probabilities. Bet365 might price CSK at 1.85 while Betfair has MI at 2.25. When you do the math, the combined implied probability drops below 100%, meaning you can back both sides across platforms and guarantee a profit.

This is called arbitrage — and it happens more often than you'd think in cricket. The catch? These windows last seconds, not minutes. You need a scanner watching odds across multiple platforms simultaneously.

So I built one. A Bloomberg Terminal-style dashboard that scans cricket odds in real-time and flags arbitrage opportunities the moment they appear.

Live site: cricketarb.com
GitHub: InoxxAIsource/cricedge-arb-scanner

The Tech Stack

I built this on Replit using their AI-assisted development tools, which let me move incredibly fast from concept to deployment.

Frontend: React + TypeScript + Tailwind CSS — dark Bloomberg Terminal aesthetic
Backend: Express.js handling odds API ingestion and serving static landing pages
ML Model: XGBoost trained on Cricsheet ball-by-ball data for independent match probability
Data: The Odds API for bookmaker odds + Polymarket for prediction market prices
Alerts: Telegram Bot API for instant signal delivery
Database: PostgreSQL via Drizzle ORM

The Arbitrage Math

The core logic is simple. For a two-outcome cricket match:

Arb% = (1/Odds_A) + (1/Odds_B)

If Arb% < 1.0 (100%), an arbitrage window exists. The profit margin is 1 - Arb%.

Example: CSK at 1.85 (Bet365) and MI at 2.25 (Betfair):

Arb% = (1/1.85) + (1/2.25)
     = 0.5405 + 0.4444
     = 0.9849 → 98.49%

That's a 1.51% guaranteed profit regardless of who wins.

What Made It Interesting: Prediction Markets

The real edge came from adding Polymarket to the scanning mix. Polymarket is a prediction market where people trade shares on match outcomes priced in cents (e.g., CSK at 52¢ = 52% implied probability).

Here's why it creates opportunities: Polymarket's trader base is crypto-native, not cricket-native. They update slower during live play compared to traditional bookmakers who have dedicated cricket trading desks. When a wicket falls in the powerplay, Bet365 adjusts in seconds — Polymarket might take 30-60 seconds. That lag is where arbitrage windows open.

The XGBoost Probability Engine

I didn't want to just compare bookmaker prices against each other. I wanted an independent probability estimate to know when ALL bookmakers were mispriced.

The XGBoost model is trained on Cricsheet ball-by-ball data and considers:

Current score and wickets
Overs completed
Run rate vs required rate
Batting team's historical performance in similar situations
Venue stats

When the model's probability diverges significantly from bookmaker consensus, it flags a value signal on top of the pure arbitrage detection.

The SEO Angle

This was also an exercise in building a content-driven product. The landing page at cricketarb.com is pure static HTML — no React shell, no client-side rendering. Google sees full content on first crawl.

I built a topic cluster architecture:

1 pillar page (3000+ word cricket arbitrage guide)
5 supporting blog posts targeting long-tail keywords
Internal links connecting everything into a tight topical web
Schema markup (SoftwareApplication, FAQPage, Article) on every page

The niche "cricket arbitrage" has near-zero competition in search. First-mover advantage is massive.

Key Learnings

Dev AI Agent is absurdly fast for going from idea to deployed product. The landing page, blog posts, and all SEO infrastructure were generated in a single session.
Static HTML > SPA for SEO. My React app returned empty HTML to crawlers until I switched the landing page to a static file served by Express.
Prediction markets are an underexplored arbitrage source. The speed gap between Polymarket and traditional bookmakers during live cricket is real and persistent.
Cricket is uniquely suited for arbitrage — high match volume (IPL has 74 matches), fast-changing in-play odds, and cross-platform opportunities that don't exist in most other sports.