Documentation

Data Dictionary & Reference

Complete definitions for every metric, term, and concept used across PredictionEdge.

Glossary

# link
Arbitrage
Simultaneously buying and selling the same market on different platforms to profit from a price discrepancy. Combined cost must be less than $1.00 for a theoretical profit opportunity.
Arbitrage Spread
Percentage profit from an arbitrage trade: ((1 - total_cost) / total_cost) x 100. A 2% spread means risking A 2% spread means risking $0.98 to guarantee $1.00..98 targeting a $1.00 return.
Divergence
A measurable price difference for the same question across platforms that may not yield profit after fees, but signals meaningful market disagreement.
Best Price
The cheapest available YES or NO share across all monitored platforms. Buying at the lowest available price may save 5-15% vs the most expensive platform.
Conviction
A composite score (0-1) reflecting how strongly traders believe in the current price, weighted by volume and liquidity depth.
Momentum
Rate and direction of price change over the most recent window. Positive = rising, Negative = falling. Scaled -1 to +1.
Confidence
Score (0-1) indicating how well two market questions from different platforms match semantically. Below 0.6 verify manually before acting.
Liquidity
Total dollar value available to trade at current prices without meaningfully moving the price.
Yes/No Price
Current market price for a YES or NO contract in cents. A YES price of 72c implies ~72% probability of the event occurring.
Implied Probability
Market-derived probability of an event, equal to the YES price. Differences across platforms create divergence opportunities.
Snapshot Bucket
A recorded sample of market metrics at a specific interval (hourly, daily, weekly). Powers trending charts and the metric trend table.

Metric Definitions

# link
MetricTypeRangeDescription
Yes Pricedecimal0.01-0.99Current bid price for a YES share. Multiply by 100 for cents.
No Pricedecimal0.01-0.99Current bid price for a NO share. Yes + No = ~1.00 in efficient markets.
Arb Spreadpercent0-20%Theoretical profit % from best arbitrage trade across platforms. Above 2% after fees is actionable.
Momentumfloat-1 to +1Price velocity. Positive = rising. Negative = falling. Magnitude = rate of change.
Convictionfloat0-1Volume-weighted certainty score. 5 dots = maximum conviction.
Confidencefloat0-1Semantic similarity between matched questions. Below 0.6 verify manually.
Divergencepercent0-100%Average cross-platform price difference. High = markets disagree = potential signal.
Best Price Deltapercent0-30%Savings available buying cheapest vs most expensive platform for this market.
Volume 24hcurrency$0+Total dollar value traded in last 24 hours. Proxy for activity and price reliability.
Liquiditycurrency$0+Total capital in order book. Higher = larger trades with less slippage.

Platform Guide

# link
PolymarketReal Money
Visit
Fees

~2%

Pros

Low fees, high liquidity

Cons

Requires crypto wallet, US restricted

KalshiReal Money
Visit
Fees

7%

Pros

CFTC regulated, USD deposits

Cons

7% fee kills small spreads

PredictItReal Money
Visit
Fees

10%+

Pros

Large political community

Cons

15% effective fee, slow settlement

ManifoldPlay Money
Visit
Fees

None

Pros

No fees, great API

Cons

No real money withdrawals

MetaculusReputation
Visit
Fees

None

Pros

High quality forecasters

Cons

No real money trading

Platform Data Limitations

# link

Each platform exposes different data through its API. Understanding these limitations helps you interpret prices and spreads accurately.

PredictIt
Multi-Contract Markets

PredictIt uses a bracket structure where one market contains multiple YES/NO contracts (e.g., "Which party wins the Senate?" has separate contracts per party). PredictionEdge displays each contract individually. Prices shown may differ from what you see on PredictIt's market overview page, which shows the full bracket.

Fee Impact

10% fee on profits + 5% withdrawal fee = ~15% effective cost. A spread must exceed ~16% to break even after fees.

Position Limits

PredictIt caps positions at $850 per contract. Large theoretical spreads may not be fully exploitable.

Polymarket
US Restricted

US residents cannot legally trade on Polymarket. Markets are monitored for divergence signals but verify jurisdictional compliance before acting.

CLOB Pricing

Polymarket uses a Central Limit Order Book. Thin markets may show wide spreads between YES and NO that do not represent true probability.

Kalshi
API Rate Limits

Kalshi's public API has stricter rate limits. During high-volume periods, price updates may lag by 1โ€“2 polling cycles (~5โ€“10 minutes).

Spread Calculation

Kalshi charges fees on both sides of a trade. PredictionEdge applies 7% per leg โ€” any Kalshi-involved spread requires >14% gross to be net positive.

Manifold Markets
Play Money Only

Manifold uses mana (แน€), a play-money currency with no real-world cash value. Use Manifold data as a sentiment signal, not a trading target.

AMM Pricing

Manifold uses an Automated Market Maker model. Prices may be less stable than order-book platforms, especially in low-activity markets.

Metaculus
Forecast Only โ€” No Trading

Metaculus is a forecasting platform with no money involved. "Prices" represent the community median probability. Useful for calibration but cannot be arbitraged.

Data Latency

Community forecasts update as users submit predictions. Data may lag real-time events by hours in fast-moving situations.

Quick Reference FAQ

# link

How often is data refreshed?

Market data is scanned every 5 minutes. The dashboard auto-refreshes every 30 seconds. Pro/Premium users receive instant push notifications for high-spread opportunities.

Why are real arbitrage opportunities so rare?

Prediction markets are efficient. Most mispricings are corrected within seconds by automated bots. Divergence opportunities (5-10/day) are the primary value driver for most traders.

Is the data real or demo?

The price chart uses stored snapshot data when available (live) or illustrative demo data when insufficient history exists. The LIVE/DEMO indicator in the status bar shows which is active.

When should I be cautious about Confidence scores?

Below 70%, manually verify questions resolve on the same event before acting. Below 50%, treat the match as a potential false positive.

Which platforms are real money?

Real money: Polymarket, Kalshi, PredictIt. Play money / reputation only: Manifold Markets, Metaculus.

Disclaimer: PredictionEdge.ai provides market intelligence for informational purposes only. Nothing constitutes financial advice. Always verify trades manually before execution.