Data & Methodology

1. Source policy

Runtime market data is obtained from official or explicitly configured sources. EventCap labels implementation capability separately from live connection status. The presence of code for an integration does not mean credentials, licenses, or fresh data are available.

2. Freshness and completeness

Timestamps, source status, collection runs, and retained snapshots are used to communicate freshness. API outages, rate limits, closed markets, missing order books, or absent credentials can reduce completeness. A blank or unavailable value is preferable to fabricated data.

3. Market matching

Cross-market relationships may be exact and manually verified, structurally similar but unverified, or candidates requiring review. Titles alone are not sufficient. Close time, resolution source, outcome meaning, eligibility, and settlement language should be compared.

4. Price gaps and execution

A raw price gap is not necessarily executable. Execution views may account for available depth, spread, fees, direction, partial fills, and quote age. Even a modeled positive edge can disappear before both legs execute.

4A. Flow Intelligence

Polymarket Flow Intelligence uses retained public proxy-wallet trades, holder snapshots and open-position samples where available. Kalshi does not expose public trader identity in the public trade feed, so EventCap reports anonymous large-trade flow only and never infers a holder or wallet. Kalshi premium notional is contract count multiplied by the public taker-outcome price. Flow Activity Index values are descriptive prioritization metrics, not probabilities or forecasts. Large-trade thresholds, observation window, coverage span and source age should be read with every flow signal. “First observed large wallet” means first observed in the retained EventCap sample for the selected window, not a newly created wallet.

5. Confidence indicators

Confidence describes source completeness, freshness, verification state, and available evidence. It is not a forecast probability and does not guarantee that a market outcome or strategy is correct.

6. Historical performance

Track-record views depend on retained signals, checkpoints, resolution status, and consistent evaluation rules. Selection effects, changing market conditions, incomplete history, and small samples can make historical metrics misleading.

T-24h calibration proximity is calculated as 100 × (1 − mean absolute probability error) over resolved markets that have an observed T-24h checkpoint. EventCap displays the qualifying sample size (N) and T-24h observation coverage alongside the value. This is a descriptive historical calibration measure, not forecast accuracy or a guarantee of future performance.

7. Closed-market historical archive

EventCap retains closed-market evidence as source observations and explicitly labeled historical aggregates rather than synthetic reconstructions. Price, quote/spread, and market-state history can be represented in durable aggregate tiers when the underlying source observations were actually collected. The production retention policy for this release keeps raw covered observations for 7 days, 1-minute aggregates for 180 days, 5-minute aggregates for 730 days, and 1-hour aggregates without automatic expiry. Pro can inspect the full per-market retention manifest and export supported retained datasets; Free can inspect the closed catalog, final snapshot, and archive-coverage preview. Missing pre-collection history is not invented or backfilled synthetically.

8. Corrections

Data defects or matching errors may be corrected as new official information becomes available. Questions or reproducible issues may be sent to support@eventcap.xyz with the market URL, timestamp, and observed discrepancy.

Active Top 100 methodology

Polymarket: Active discovery uses official open, order-accepting markets with rolling volume24hr of at least $100,000. Individual markets are ranked by current 24-hour USD volume. Events are ranked by the sum of qualifying child-market 24-hour volume and are excluded when no qualifying child remains. Historical cumulative volume is context only and never controls the active rank.

Kalshi: Active discovery uses official markets returned by the open-market feed whose raw state normalizes to canonical OPEN_TRADABLE (open or observed active) with positive current volume_24h_fp. Individual markets and grouped events are ranked by rolling 24-hour contract volume. Units remain contracts; EventCap does not infer a USD cash-volume equivalent.

The lists contain up to 100 qualifying rows. Dead or low-activity markets are never used as padding. Historical observations remain available through retained research and archive workflows rather than being deleted.