Data interpretation
Monitored volume is not exchange-wide volume.
Aggregates become misleading when their denominator disappears. This note defines the difference between volume in Polytrack's monitored records and total Polymarket activity, then gives a checklist for reading rankings and trends consistently.
Every aggregate needs a population and a window.
Polytrack analytics begin with stored monitored trades that match a selected time range. From that set, the product can sum trade value, count records, count distinct wallet addresses, group categories or recorded sides, and rank markets or wallets. The correct population is therefore “records available in Polytrack for this window,” not “all activity on Polymarket.”
Removing the coverage phrase silently changes the claim. A chart labeled $1 million of tracked volume is not evidence that the exchange traded only $1 million, nor that Polytrack captured a fixed percentage of the total.
Buckets describe when records entered the selected series.
Time-series panels group records into bucket sizes chosen for readability. A short range can use smaller buckets; a long range can use larger ones. Late or corrected records can alter a historical bucket, and timezone or boundary choices can change which bucket receives an event.
A rising bucket shows more monitored value in that portion of the selected range. It does not by itself show causation, net exposure, liquidity, or future price direction.
Ranks inherit the same coverage limit.
The Polytrack leaderboard groups monitored trades by wallet and builds a leading candidate set by tracked volume. It can then show volume and average-trade-size lists from that set. A wallet outside the monitored data or outside the candidate set cannot appear, even if its broader exchange activity is large.
Average size also depends on the monitored trade count. The figure should be read with count, window, and largest tracked trade—not as an exchange-wide characteristic of the wallet.
- Keep tracked or monitored in the metric label.
- Keep the selected range beside the value or rank.
- Use trade count and coverage notes to qualify averages.
- Do not translate an activity rank into a performance rank.
Comparisons are valid only after scope alignment.
Two Polytrack panels can be compared when they use the same monitored population, filters, value definition, and time boundary. A Polytrack total and an external exchange-wide number are not directly comparable until the external source, included event types, time zone, corrections, and denominator are reconciled.
When that reconciliation is unavailable, the honest conclusion is limited: Polytrack observed a pattern in its monitored data. That statement can still be useful and reproducible without pretending to describe the whole exchange.
Limitations
Keep these with the conclusion.
- The monitored dataset can contain ingestion gaps, late records, or later corrections.
- Trade value is a descriptive notional field, not open interest, liquidity, or net exposure.
- Distinct addresses are not necessarily distinct people or organizations.
- Category and new-wallet fields are classifications whose quality depends on available context.
Sources