VIGIL is a research preview. We read public Polymarket data and public on-chain state, and we publish a skill grade — A through F — for every forecaster we can identify. The grade comes with a bootstrap 95% confidence interval. If the interval is wide or the sample is small, we show INS (Insufficient Data) instead of a letter.
We never hold user funds. We never sign transactions. We never request private keys.
VIGIL does not place trades, mirror positions, or act on any user's behalf. Grades are informational.
We pull from Polymarket's public Gamma API, data-api, CLOB log, and on-chain USDC/redemption events. No private data.
The free tier has no login. Pro and Elite use simple email-based API keys. No KYC.
We have read Polymarket's Terms of Service carefully. VIGIL reads publicly exposed data through stable, documented endpoints; we observe all published rate limits; we do not attempt to re-identify or deanonymize traders beyond what Polymarket itself publishes via handles and public profiles.
We believe independent skill analytics improve prediction market quality — calibrated traders get capital from followers, poorly calibrated whales lose influence, consensus surfaces real information. If Polymarket believes otherwise and reaches out, we will engage in good faith. In the interim, everything we publish is reproducible from public inputs, and every grade links to the exact data used to compute it.
"Research preview" is not marketing language. It means three things:
1. Grades may change as our method improves. We log every methodology revision in the public changelog. A wallet's grade today may differ from its grade tomorrow if we fix a bug or widen our sample. We version the formula explicitly (currently v1.22.1).
2. We publish errors. If you find a mistake — in the math, the code, or a specific grade — email api@vigilscore.xyz or open a GitHub issue. We will credit you publicly and fix it.
3. Not investment advice. A VIGIL grade is a historical calibration statistic, not a trade signal. Skilled traders can be wrong; new information moves markets before it moves grades. Do not copy-trade based on a grade. Use it as one input among many.
We persist three things: (a) wallet addresses we have scanned, (b) the grades and CIs we have computed, (c) email addresses of alert subscribers who explicitly opted in. Nothing else.
If you operate a wallet and want it removed from the leaderboard (but not from opportunistic single-wallet scoring, which is unavoidable because the underlying data is public), email api@vigilscore.xyz from an address you can prove controls the wallet (signed message), and we will exclude it from all aggregated surfaces within 72 hours.
Prediction markets are an information-aggregation technology. They work better when the information is good. Today, the loudest wallets on Polymarket are not the most accurate; followers copy them; capital flows the wrong way; markets are noisier than they should be. VIGIL is the tool that fixes the attribution gap — the bridge between "this wallet has PnL" and "this wallet has skill." The formula is public. The code is inspectable. The moat is the pipeline, not the secret.