Live Edge / Mispricing Board Developer

CrazyLive (开心直播) produces 错价时刻 (Mispricing Moment), a live prediction-markets and sports-odds show that airs five nights a week (Mon–Fri), starting 7:00 PM U.S. Pacific Time. The heart of the show is a live mispricing board: whenever Polymarket, Kalshi, Vegas/sportsbook lines, and our own signal disagree by a meaningful margin, that divergence is the edge — and we want it detected and highlighted on screen, live, automatically. We're looking for someone who can both build the data/analytics pipeline and operate it during the show. What we want on screen: A live board that converts every source to implied probability so they're apples-to-apples Auto-computed divergence / spread between sources for each event A "错价时刻 / Mispricing Moment" highlight that triggers automatically when divergence crosses a set threshold (color flag, sort the biggest edges to the top) All numbers auto-refresh during the broadcast — not manually retyped Deliverables: Build the pipeline (mostly one-time): Pull live prices via API — Polymarket API, Kalshi API, and a sports-odds API (e.g. The Odds API). Normalize every source to true implied probability — including removing the vig/overround from sportsbook odds so the comparison is fair. Compute cross-source divergence, apply the threshold/alert logic, and push it to auto-updating on-screen titles. Target setup is vMix Data Sources (preferred) or an OBS browser-source overlay. Operate it live each show: Curate which events appear, watch for mispricing moments as they trigger, and surface them on cue. Step in immediately if a feed drops. Requirements: Strong grasp of implied probability and odds math — converting American/decimal odds, and de-vigging sportsbook lines to true probability. This is the core skill; a board that doesn't de-vig shows fake edges. Reads a prediction-market price (Polymarket/Kalshi) as a probability Comfortable with REST APIs + Google Sheets and/or light scripting (Python or JS) to compute the divergence and feed it on-screen Experience building live data overlays for vMix (Data Sources) or OBS (browser source) Reliable during U.S. Pacific-Time show hours; calm under live conditions Nice to have: sports-betting, arbitrage, trading, or fintech-data experience Schedule: Pipeline build first, then live operation five nights a week (Mon–Fri) — show starts 7:00 PM U.S. Pacific Time, ~[60–90] min per show This is a recurring, ongoing commitment: you must be reliably available every weekday evening Pacific Time. Having a backup arrangement for nights you're out is a strong plus. Extra one-off broadcasts during campaigns (e.g. our World Cup series), with notice To apply, please include: Examples of live data overlays, odds tools, or dashboards you've built (links if possible) Whether you'd target vMix Data Sources or an OBS overlay, and why Your experience with odds math / prediction-market data (Polymarket, Kalshi, odds APIs, de-vigging) Your pricing: a fixed price for the pipeline build + a per-show or hourly rate for live operation