# A quote is not a fill: an order-size audit
Only 63.6% of eligible Polymarket books captured enough asks for 10,000 contracts. We tested 100, 1,000 and 10,000 contracts in a dated, selected watchlist.
Canonical URL: https://vultax.com/research/prediction-market-order-size-depth-audit
Author: [Vultax Research](https://vultax.com/editorial-policy)
Published: 2026-09-20
Last public revision: 2026-09-20
Published research snapshots and calculations; these exports do not contain current quotes or live account data. Observation windows and populations are stated in the source captions and methodology. Missing metadata is unknown, not zero. Hypothetical static book walks on selected, unmatched watchlists. Captured depth is not guaranteed liquidity or actual execution. Fees and delays are excluded. No claim of achievable profit or platform-wide fill rates.
## Key figures
- Books sampled: 10,282 — First recorded cycle each UTC hour, 17–18 September 2026; selected watchlists.
- Polymarket depth: 1,717/2,701 — Captured enough asks for 10,000 contracts in the primary subset.
- Kalshi depth: 680/1,188 — Separate Yes-book subset; contracts are not matched across venues.
## The short answer
A copy-trading screenshot can show a price without showing how much was available at it. Vultax walked the captured ask levels in 10,282 book snapshots at three fixed sizes. In the primary Polymarket subset, all 2,701 books contained enough asks for 100 contracts, but only 1,717 did for 10,000. In the separate Kalshi subset, the corresponding counts were 1,187 of 1,188 and 680 of 1,188.
These are displayed-capacity checks, not observed fill rates. A missing quantity can exist beyond the 20 captured levels. A displayed order can disappear before a trader reaches it. The samples cover different watchlists, so the figures do not rank one venue against the other.
## Size changes the denominator
We keep books with a positive, uncrossed bid and a best ask between 10 and 90 cents. This prevents the most extreme one-sided books from dominating the primary cost comparison. Polymarket contributes token-side books; Kalshi contributes Yes books reconstructed from No bids. The row is a snapshot, not an independent market or trader.
### Size changes the denominator
Extra cost is ask-walk VWAP minus best ask, among completed hypothetical purchases only. Contracts are quantities, not dollars. No fees.
| Venue subset | Contracts | Enough captured depth | Share | Median extra cost, cents/contract |
| --- | --- | --- | --- | --- |
| Polymarket | 100 | 2,701/2,701 | 100.0% | 0.000 |
| Polymarket | 1,000 | 2,111/2,701 | 78.2% | 0.000 |
| Polymarket | 10,000 | 1,717/2,701 | 63.6% | 0.250 |
| Kalshi | 100 | 1,187/1,188 | 99.9% | 0.000 |
| Kalshi | 1,000 | 1,140/1,188 | 96.0% | 3.487 |
| Kalshi | 10,000 | 680/1,188 | 57.2% | 15.381 |
[CSV](https://vultax.com/research/prediction-market-order-size-depth-audit/data.csv?dataset=capacity-table) · [JSON](https://vultax.com/research/prediction-market-order-size-depth-audit/data.json?dataset=capacity-table)
### Polymarket: enough depth in captured asks
Denominator: 2,701 eligible token-book snapshots; 20 captured levels, no actual orders.
| Category | % (%) |
| --- | --- |
| 100 contracts | 100 |
| 1,000 contracts | 78.15623843021103 |
| 10,000 contracts | 63.56904850055535 |
[CSV](https://vultax.com/research/prediction-market-order-size-depth-audit/data.csv?dataset=capacity-chart) · [JSON](https://vultax.com/research/prediction-market-order-size-depth-audit/data.json?dataset=capacity-chart)
## A larger order can appear cheaper when thin books disappear
The Polymarket 90th-percentile extra cost falls from 9.381 cents at 1,000 contracts to 1.649 cents at 10,000 contracts among completed walks. That is a selection effect in the comparison: the completion set changes. It does not show that increasing order size improves execution in the same book.
For a fair size curve, hold the book set fixed. Among the 1,717 Polymarket books that can complete every size, median extra cost is 0, 0 and 0.250 cents for 100, 1,000 and 10,000 contracts. Among the 680 common Kalshi books, the figures are 0, 1.710 and 15.381 cents. Every individual completed walk has nondecreasing average price as size grows; the reproduction checks this invariant.
Conditional medians can still hide costly tails. A zero median means at least half of those observed walks stayed at the best ask, not that a user can safely expect zero slippage.
## How the replay works
From the closed 17 and 18 September 2026 archive files, select the earliest recorded book cycle in each UTC hour on each venue. The resulting records cover 48 cycles, 83 distinct Polymarket markets and 407 Kalshi tickers, including rolling contracts. The archive contained 616,950 book records before hourly selection. All sampled records reported successful collection, but empty sides remain empty.
For an order of Q contracts, sort asks by price, consume the available quantity at each level, and divide the sum of price times quantity by Q. If the captured levels cannot supply Q, mark the walk incomplete and leave its cost missing. Never impute the last visible price for missing depth. No capital was deployed.
The watchlist was assembled for event research, not drawn randomly. Repeated hourly books and complementary token sides are dependent. The study excludes execution fees, copy-service fees, transfer costs, order minimums, market impact response, competing orders and the delay between seeing and submitting an order. Those omissions make this a first execution check, not a profit backtest.
## Check the calculation
[displayed-execution.csv](https://vultax.com/datasets/prediction-market-order-size-depth-audit/displayed-execution.csv) · [results.json](https://vultax.com/datasets/prediction-market-order-size-depth-audit/results.json) · [METHODS.md](https://vultax.com/datasets/prediction-market-order-size-depth-audit/METHODS.md) · [Reproduction bundle](https://vultax.com/datasets/prediction-market-order-size-depth-audit/reproduction.zip). The bundle contains the calculation, frozen analysis inputs, results and a SHA-256 manifest.
Reuse the figures with a link to this study and the stated population and observation window. Vultax welcomes corrections and independent methods reviews. A citation or collaboration does not require a positive conclusion or a reciprocal link.
## What a copy-trading partner could add
A useful joint follow-up would match an alert's original timestamp to the first observable follower quote, then report 100, 1,000 and 10,000-contract walks with explicit fee schedules. The partner can contribute an anonymized alert log and its execution rules; Vultax can produce the replay and failure table. Failed, late and incomplete observations belong in the published denominator alongside the appealing examples.
## Sources and methodology
- [Polymarket public order-book schema](https://docs.polymarket.com/api-reference/market-data/get-order-book) — Price and size levels; read 19 September 2026.
- [Kalshi order-book API](https://docs.kalshi.com/api-reference/market/get-market-orderbook) — Yes and No bid books; asks derived from the opposite bid.
- [Vultax copying-cost study](https://vultax.com/research/polymarket-copy-trading-costs-and-tools) — Related historical comparison, with a different dataset and question.