How Polyguana Ranks Prediction Markets — Scoring Methodology

Volume is the fastest way to tell whether a prediction market is attracting real attention. A market with steady trading is harder for a single participant to move, more likely to incorporate fresh information, and generally more informative than a market with only a few isolated fills. For that reason, Polyguana treats turnover as a primary ranking input rather than a vanity metric.

The volume lens is intentionally front-loaded toward recent activity. Twenty-four hour volume carries the most decision weight because it reflects what traders are acting on now. Seven-day volume acts as a stabilizer so one sudden burst does not automatically outrank a market with consistent engagement. Thirty-day volume adds context by separating markets with durable participation from markets that only flashed briefly. In practice, that means the score rewards both immediacy and persistence, but it never lets an old liquid market coast forever on history alone.

That time-frame mix matters because prediction markets behave differently around catalysts. Election nights, court rulings, earnings calls, and geopolitical headlines can produce abrupt surges in trading. A purely short-term ranking would overreact to those bursts. A purely long-term ranking would miss the point of real-time markets altogether. The blended volume view is designed to keep both truths in frame at once.

Raw volume is not enough if the market is expensive to trade. Liquidity measures whether someone can actually enter or exit near the displayed price. Polyguana looks at two closely related pieces of execution quality: the bid-ask spread and the available depth near the midpoint. Tight spreads mean the quoted probability is more actionable. Deeper books mean larger trades can happen without moving the market materially.

Thin liquidity is a bad signal because the displayed price can look precise while still being fragile. A one-cent move in a deep market is meaningful; a one-cent move in a thin market may just reflect a single small order. That is why the platform also exposes a quality-style precision score built from spread, depth, and freshness. Markets with poor execution quality can still be listed, but they are less likely to be treated as top-tier forecasting signals.

At the product level, that execution score is intentionally mechanical. Spread is weighted most heavily, because a wide quote immediately degrades trust. Depth and freshness matter next, because stale or shallow markets can look healthy until you try to trade them. This is the same reason Trader mode defaults toward execution-sensitive metrics: the best-looking probability is not very useful if it is expensive or impossible to access in size.

Probability itself also carries information. Markets near 50/50 are usually where uncertainty is still highest, disagreement is still active, and new information can matter most. Markets pinned at 98% or 2% can be important, but they often contribute less new forecasting insight because most of the debate has already been resolved. Polyguana therefore gives more analytical weight to markets that remain genuinely contested.

This does not mean near-even markets automatically rank above everything else. Instead, probability is treated as one dimension of informational value. A contested market with weak liquidity or tiny volume can still be noisy. A heavily traded 80/20 market can still matter if new developments are forcing repricing. The goal is to identify where probability is both meaningful and supported by enough market structure to trust what it is saying.

Not every market is equally well designed. Resolution criteria need to be clear, verifiable, and resistant to subjective interpretation. Markets with vague wording, ambiguous deadlines, or questionable source language can trade actively and still produce weak forecasting value. Polyguana treats quality as a distinct layer because good ranking should reward clean contract design, not just trading intensity.

Quality also includes practical trust signals: how consistently similar markets have resolved, whether the event structure is easy to follow, and whether the market is likely to stay tradable through the period when information matters most. Analyst-oriented views lean harder on this structural lens, because the point is not simply to find motion, but to compare markets whose prices deserve deeper interpretation.

In practical terms, quality is the part of the ranking that keeps the product from becoming a pure momentum board. A badly worded market can still show a dramatic price move or impressive turnover. That does not make it a good forecasting instrument. The methodology deliberately pushes back on that kind of false prominence.

Polyguana does not rely on one universal black-box score. It normalizes the major inputs onto comparable ranges, then combines them in weighted form depending on the surface. Pulse leans more heavily on timeliness and narrative movement. Trader leans harder on recent volume, liquidity, spread, and execution quality. Analyst shifts more weight toward probability context, quality, and resolution structure. The common principle is that each input must earn its place in the final order.

When a blended ranking is needed, recent activity receives the strongest influence, liquidity and quality act as trust multipliers, and probability context helps decide whether a market is informative or already settled. User-selected sort still matters: if you sort by liquidity, resolution time, or change, that explicit choice overrides the default emphasis. The methodology is therefore opinionated but inspectable, which is exactly how ranking products should behave.

This also means rank position is not just a popularity contest. A market can lose rank because volume dried up, because the spread widened, because the probability is no longer contested, or because the contract design makes interpretation weaker than it first appears. Composite ranking is meant to reflect usability, not just attention.

Prediction markets are valuable because they turn disagreement into probabilities. The usefulness of those probabilities depends on calibration: a market priced at 70% should resolve true roughly seven times out of ten over a large enough sample. In forecasting research, Brier score is one common way to judge that calibration by penalizing confident errors more than cautious ones. Good market ranking helps users focus on contracts where that calibration is more likely to mean something.

In other words, the methodology is not about making the page look busy. It is about reducing noise. By favoring active, liquid, interpretable markets over stale or structurally weak ones, Polyguana aims to surface markets that are better candidates for real analysis, faster monitoring, and better forecasting decisions.

That is the practical job of a ranking system in forecasting: reduce the amount of time users spend filtering out bad data. If the page consistently elevates markets with better information density and lower execution friction, it becomes more useful for monitoring fast-moving events, comparing alternative narratives, and building a better mental model of what the market really believes.