Methodology — How We Research Polymarket Markets…

Why we publish our methodology

Polymarket prices look like probabilities, but only if the underlying market is liquid, well-defined, and resolves cleanly. Our research methodology exists so readers can verify how each storyline, guide, and news brief was built, and decide independently whether the analysis is credible.

Market selection

We focus on markets that meet at least three of the following:

$100K+ in cumulative volume or $10K+ in 24h volume.
Order book depth of at least $5K within 5 cents of mid-price on both sides.
Clearly written resolution rules that reference a deterministic source.
Material relevance to an Israeli or globally-followed news cycle.
At least 7 days remaining before resolution (so analysis has shelf-life).

Markets that fail all five criteria may still appear in news briefs but never as the centerpiece of a storyline.

Data sources

On-chain (Polygon)

Polymarket CLOB contract event logs for trade history, order book depth, and volume.
Conditional Tokens Framework (CTF) contracts for outcome share supply and redemptions.
USDC.e / pUSD balance tracking for treasury and trader positions.
UMA Optimistic Oracle contract for proposed and disputed resolutions.

Off-chain

Polymarket gamma API for market metadata (questions, end dates, categories).
Polymarket CLOB REST/WS APIs for current order book and trades.
Tier-1 news outlets and government publications for fact-base and context.
Live-search verification of claims via Grok / Perplexity / Gemini grounding.

How we turn prices into probabilities

The midpoint of the YES/NO order book is our headline implied probability. We adjust for:

Spread: if bid-ask is wider than 4 cents, we report the bid-ask range instead of mid.
Liquidity: if executing $1K size moves price by more than 2 cents, we flag the market as illiquid and weight the signal lower.
Resolution risk: for markets where the rules are ambiguous or could go to a UMA dispute, we report a confidence band around the price (e.g. “market implies 62%, our estimate is 50–70% accounting for resolution risk”).

UMA monitoring

Every market we cover has its UMA proposal/dispute history checked. If a market has been disputed in the past or is similar to one that was, we note this in the storyline. We never publish a definitive resolution call until UMA has finalized — even if Polymarket's UI shows the outcome.

News verification

For event-driven markets we cross-check news in three places before treating an item as resolution-relevant:

The original primary source (official statement, agency wire, government release).
At least one independent tier-1 outlet's reporting.
The market's specific resolution rules to confirm the news actually triggers a YES/NO outcome.

Items that pass only step one are flagged as “reported, not yet resolution-relevant”.

Update cadence

Storylines: reviewed at least weekly while the market is open; updated immediately on any UMA proposal or material news.
Guides: full re-review every 90 days; immediate updates when Polymarket platform mechanics change (e.g. the April 2026 USDC.e → pUSD migration).
News briefs: snapshot at publication; corrections only as new facts emerge. We don't silently re-time-stamp.
FAQ: reviewed monthly; entries dated.

Tools and code

Our analysis tooling includes Python scripts for on-chain queries (web3.py / ethers.js), Grok and Claude for first-draft research, and a custom storyline pipeline that ingests UMA events and Polygon trade logs. Code that's safe to share publicly will be open-sourced as we iterate.