Follow the smart money. A real-time research dashboard for public prediction markets — wallet tracking, consensus, value reads, and alerts, in one clean feed.
Research only · Not financial advice · Installs on any phone & desktop
Prediction Command Center watches public prediction markets in real time — tracking which sharp wallets are moving, where independent signals agree, and which markets look mispriced versus the models. It reads the market for you and shows the state of each position; it never places a bet, holds funds, or tells you what to do. Every call it makes is logged and graded, wins and losses alike, so the track record is public.
Follow which tracked wallets are buying and selling, and how much is behind each move.
Every live signal grouped by game, hottest first — the whole board at a glance.
Where independent reads (news, sharps, tape) line up on the same market.
Markets that look underpriced or overpriced versus the models — described, not prescribed.
Continuously updated totals and prices, with opt-in per-sport alerts capped so they never spam.
Every call scored and kept — a public, honest scorecard of hits and misses.
A solo project that grew from a local script into a public, installable, cross-platform app — without a cloud host or a native rewrite. The engine stays private; the delivery stack is below.
A Python system scans public markets, tracks smart-money wallets, scores signals, and renders a live HTML command center — running on a home PC.
A Preact front-end plus a web manifest and service worker made it installable and offline-capable on any device.
A named Cloudflare Tunnel exposes the local app at a permanent HTTPS domain — no server, no hosting bill.
PWABuilder wrapped the same PWA into a Microsoft Store MSIX, a Google Play Android bundle, and an iOS project — one app, three stores.
An Azure instance keeps grading the track record 24/7, and everything auto-syncs to a private GitHub repo.
I documented the whole store-shipping process as a free, open-source playbook so others don't have to piece it together the way I did → the build guide.
This started as a personal research project and grew, version by version, into a public, always-on, installable app — built solo, part-time. Here's the honest timeline of how it got here.
The core idea: in prediction markets like Polymarket and Kalshi, every trader's track record is public. So instead of guessing, find the traders who've been provably good for a long time, watch their money move in real time, cross-check it against real bookmaker odds and fresh news, and put it all on one live page. It grades its own calls against reality and never places a bet, holds money, or touches an account.
v0.1 could only watch the markets while the home PC was awake. v0.2 moved the engine to a dedicated always-on cloud server that runs around the clock — scanning, tracking, and grading whether the computer at home is on, asleep, or unplugged.
The final leap: turning a tool that lived behind a private link into something anyone can install — without a native rewrite or a cloud host for the front-end.
A quick, no-jargon glossary for the pieces that make this work — so the "how" is clear even if you're not a developer.
An exchange (Polymarket, Kalshi) where people trade on the outcomes of real events. Prices act like live odds, and — unlike a casino — every trader's position and track record is public.
A "wallet" is one trader's public account. "Smart money" is the handful whose history proves they've been right over time — the ones worth watching when their money moves.
A doorway one program uses to ask another for data. The agent uses APIs to pull market prices, trade activity, sportsbook odds, and news — automatically, every few seconds.
A large language model — the kind of AI behind tools like Claude. Here it's used to read fresh news in plain language and turn it into a probability the math can weigh, not to make decisions on its own.
A program that runs on its own, on a loop: gather data → weigh it → surface what matters → check itself — over and over, without someone driving each step. That's what "the agent" is.
A Progressive Web App — a website built so it can be installed like a real app, work offline, and live on your home screen. It's what lets one codebase ship to every phone and desktop.
A secure pipe that gives an app running on a home computer a stable public web address — so the world can reach it without renting a server.
After each game resolves, the agent marks its own earlier call a win or a loss and keeps the record forever. If an idea stops working, the scorecard says so — no cherry-picking.