Splitrule

The Statistics Page. Every figure with its source, its capture and its date.

FEE TRUTH · THE LEDGER

Prediction Market Statistics 2026: What Does Trading Actually Cost?

Splitrule's figures ledger as a public product: every number states where it came from and when it was read.

What does the fee schedule say? (verbatim)

Polymarket US, from its published schedule, effective July 1, 2026, captured 2026-08-04:

taker:  0.06 x C x P x (1-P)          1.50¢ per contract at 50¢
maker:  rebate of 0.0125 x C x P x (1-P)

The terms above, their two clocks and the daily check on them are on the venue schedule panel, and the edge a trade needs at each price is the panel below it.

What gross edge do you need to break even? (derived)

Taker on Polymarket US costs 0.06 × P × (1−P) per contract, so the edge one leg has to clear is that number and the price decides it. Formula and grid, in cents per contract:

price   taker, one leg   taker, round trip
10¢     0.54¢            1.08¢
30¢     1.26¢            2.52¢
50¢     1.50¢            3.00¢
70¢     1.26¢            2.52¢
90¢     0.54¢            1.08¢

The takeaway. A taker round trip at 50¢ costs 3.00 points of edge on this venue, and the fee is largest at 50¢ and smallest at the edges. The per-price arithmetic and its receipts are on the venue fee panel.

How is the Polymarket US catalogue structured? (measured)

One complete walk of the open catalogue: 15,508 open markets, captured 2026-07-25 (capture G1, population digest a7ef2cee…). The catalogue moves. These figures describe that capture, not today, and no two figures from different captures are differenced on this page.

tick size   markets   share of open catalogue
$0.01       14,372    92.67%
$0.001      769       4.96%
$0.005      367       2.37%
total       15,508    100.00%

The takeaway. 1,136 of 15,508 open markets (7.33%) tick finer than a cent. Code that assumes a penny grid mis-prices them. The fee coefficient was a single value, 0.06, across every open row of the capture, which corroborates the published schedule rather than replacing it.

How does Splitrule verify a number? (methodology)

Fee quotes are extracted from saved copies of the venues' own pages, hashed at capture time, and a standing test asserts that every quote on this site appears byte-for-byte in its named capture. Before a schedule ships, a human compares the quoted constants against the live page once and logs the dated confirmation. Numbers from trade press or search summaries never ship, and two figures from different captures are never quoted together.

The takeaway. If a figure has no receipt, it does not ship.

Where does the open interest sit? (reported)

The category table that stood here is a panel of the ladder now, beside the expiry counts it is a table twin of, and it prints the same rows from the same pin: open interest by category. This heading keeps its address so a reader who saved it still reaches the figure.

How often does a venue edit its own rules text? (measured)

40 intervals of Polymarket US rules text, 2026-07-31 to 2026-09-08, joined from 41 usable captures. The rate below counts only the 32 intervals at least 20 hours wide and measured end to end by one instrument, because an all-interval rate would mix 0.01-hour and 48.01-hour intervals under one denominator.

carried market-intervals   727,579
changed                   386
day-scale change rate     0.0531%

The takeaway. Deadlines moved, rules text changed length bracket and wording changed. Not one market changed the source its outcome is read from or changed its void language. A market counts here only when it is carried through both captures of an interval, so this is a lower bound on what changed and never an estimate of it. The axis, the class mix and every receipt are on the rules census on the venue screen.

The instrument-side series (download)

This desk publishes its own measurements of its own instruments as a versioned file: how many capture runs it made, how many its completeness guard refused, over what span and at what cadence, how many watches it declares, and the digest of the artifact each figure is read from. Every figure names its instrument and the path it is read from, so a reader can re-derive it rather than trust it.

Download the series (JSON)

Cite this series by its DOI: https://doi.org/10.5281/zenodo.22240062

Licence: CC BY 4.0 (CC-BY-4.0) — https://creativecommons.org/licenses/by/4.0/

This release is Splitrule's own measurements of its own instruments: counts of its runs, refusals and watches, the cadence at which they fired, and digests of the artifacts it produced. It contains no venue market data and no material derived from any venue's market data.

The licence is offered over whatever rights Splitrule holds in this compilation — its selection, arrangement, structure and accompanying text. It is not a claim of ownership over the underlying facts. Individual facts are not ours to license, and where a use needs no permission, none is required here.

What we ask, and what we require, are different things. We ask that you cite this release by its DOI, because that is how a measurement series stays traceable to the method that produced it. Where CC BY applies, that attribution is a condition of the licence. Where it does not, the request stands on its own and nothing in this notice makes it more than a request.