VANTEDGE POLY

How it works

An AI that prices Polymarket - and proves it.

Most tools either track whales or guess with AI. Vantedge Poly does both, and keeps a public score. Here's the pipeline, end to end.

01

Read every market

We pull the full set of active Polymarket questions with live prices, volume, and category - the same data the crowd trades on.

02

Debate both sides

For a market you forecast, an AI runs a bull case and a bear case over live web evidence (and on-chain data for crypto), then a judge weighs them.

03

Estimate the true probability

The judge outputs a calibrated P(YES) - the model's honest estimate of the real chance the answer is YES, independent of the market price.

04

Find the edge

We compare that P(YES) to the live price. A gap is an edge: the crowd may be mispricing it. ▲ YES or ▼ NO, sized by how big the gap is.

05

Check the smart money

We fuse the model with the whale tape and top-holder concentration. The ⚡ convergence signal fires only when the AI and the smart money agree.

06

Keep score, publicly

Every forecast is logged before it's sized, then scored by Brier against resolved markets. The Model-record rail shows whether the AI is actually right.

Where the data comes from

Polymarket Gamma APImarkets, prices, volume, resolution
Polymarket Data APIwhale trades, top holders, wallet P&L
Web search (Serper → SearXNG → DDG)live evidence for the bull/bear debate, one shared source
Hyperliquidlive mark / funding / OI for crypto questions

How the “right bet” suggestion works

A forecast is only half the job - the other half is how much to bet, if at all. The decision panel turns an edge into a sized, paper-only suggestion using a disciplined, cautious method:

Win probability

Our calibrated P(YES) for the side worth backing - not the raw model number, but one nudged toward how the model has actually scored on resolved markets.

Edge vs price

We only suggest a side when our fair value sits at least ~5 points off the live price. Inside that band it's a NO BET - the crowd is close enough to right.

Fractional Kelly

Kelly maths says the mathematically optimal stake for an edge; we deliberately bet a FRACTION of it (quarter-Kelly by default), scaled by conviction tier and shrunk further when the model has few resolved bets to prove itself.

Safety caps

The stake is capped at a small percentage of your paper cash, trimmed during a drawdown, and limited by how much you already have riding on correlated markets. Sizing errs small on purpose.

Order-book check

Beside the suggestion we show live bids, spread, and 24h volume, so a 'great edge' in an illiquid market reads as the trap it usually is.

Everything here is paper (simulated). The suggestion is a research aid, never advice - you decide what, if anything, to do on Polymarket yourself.

Smart Score (wallet grading)

On the Grade and Smart Money pages, each wallet gets a Smart Score - risk-adjusted skill, not raw profit. It rewards traders who win consistently and at honest odds, shrinks toward zero for small samples (so a lucky 3-bet wallet can't look elite), fades stale open positions, and credits genuine entry edge - buying meaningfully below where our model later priced the market.

Run by us, free for you

Every forecast runs on our central engine - no key to bring, no signup, no metering. The forecaster's prompts stay server-side, so you get the answer, not the recipe. And we publish every forecast we run to the public track record.

See today's edge

Open the markets

Research & simulation only - not financial advice. In-app trading is paper (simulated). Not affiliated with Polymarket.