18+ | Gamble Responsibly | BeGambleAware.org | T&Cs Apply
V
ViktoryBet
Strategy

Kalshi Contract Terms 2026: 90% Are Shared

Kalshi contract terms look per-market. We read all 16,761 series: 3,245 documents, 90% shared, and all 30 PDFs we opened still have blank placeholders.

Affiliate disclosure: contains affiliate links; we may earn a commission at no extra cost to you. We have no commercial relationship with Kalshi. 18+ only — see BeGambleAware.org. Last checked November 2026.

Key takeaways

  • The Kalshi contract terms document is a family template, not your market's contract. We read all 16,761 series in one response: they resolve to just 3,245 distinct documents. 15,083 series — 90.0% — share theirs with at least one other market.
  • One PDF covers 493 markets. ELECTION.pdf is the contract terms for 493 series. 22 documents serve 100 or more series each, covering 4,229 between them. Only 1,678 series (10.0%) have a document of their own.
  • All 30 documents we opened still contain unfilled merge tokens. Not a handful — 30 of 30. The widely-cited advice that the PDF is the authoritative text “where the API has a placeholder” is wrong: the PDF has the same placeholders, and usually more.
  • The same document is published under four different names. Checksumming our 30 samples gave 26 distinct documents, with ENTITYOUTCOME, NASCARCHALLENGE, NFLDIVISIONTOTALWINS and WCFASTESTGOAL all serving identical text.
  • This is a documentation gap, not a missing rule. Your market's actual parameters are published — as structured API fields, not as prose. rules_primary was empty on 7,985 of 8,000 open markets we checked, so no document anywhere combines the template with your market's values.

Every guide to reading a prediction market tells you the same thing: the summary on the market card is marketing, so open the Kalshi contract terms PDF and read the real rules. It is good advice with one problem nobody has checked — that PDF is usually not about your market.

The stakes are the whole position. If you are trading a mention market on a specific word, or an election market on a specific seat, the document you were told is authoritative may be the same file served to 205 other markets, with your word and your seat left as a blank in angle brackets.

So we counted. One request to Kalshi's public series endpoint on 20 November 2026 returned all 16,761 series as complete objects, and those 16,761 markets point at just 3,245 distinct documents. Then we downloaded 30 of those PDFs, extracted the text and checksummed it.

We have run this field before and got it wrong. Our own prohibited-traders census told readers the contract terms PDF is “authoritative where the API text has an unfilled placeholder”. It is not, and the correction is below.

What the Kalshi contract terms field actually points at

Kalshi groups markets into series, and the legal documents attach at series level — the same level as the fee coefficient and the prohibited-trader list. Two fields carry documents. Kalshi's API reference defines them in the singular and marks both required:

  • contract_terms_url — “ContractTermsUrl is the URL to the current terms of the contract underlying the series.”
  • contract_url — “ContractUrl provides a direct link to the original filing of the contract which underlies the series.”

Read that as a trader and you expect a document about the contract you are buying. Nothing on the page says the URL may be shared, because nothing on the page addresses uniqueness at all. That silence is the entire finding.

Leave the optional category parameter off the series endpoint and it returns the whole board in one response, which makes this a census rather than a sample. All 16,761 series carry a contract_terms_url; none is missing. But the URLs collapse hard: 3,245 distinct values, a mean of 5.2 series per document.

Contract terms documentSeries sharing it
ELECTION.pdf493
ECONSTAT.pdf255
SOCCERGAMEWIN.pdf246
SOCCERWHOSTAT.pdf222
EARNINGSMENTION.pdf218
RT.pdf207
MENTION.pdf206
SOCCERSPREADS.pdf205
POLITICALSTAT.pdf204
REPORTTOPIC.pdf192

Those ten documents are the contract terms for 2,448 markets. Widen it to every document serving 100 series or more and 22 files cover 4,229 series. The naming gives it away once you look: SOCCERGAMEWIN, POLITICALSTAT, REPORTTOPIC are product families, not markets.

90% of Kalshi markets have no document of their own

The headline number is the inverse of the sharing. Of 3,245 documents, 1,678 serve exactly one series — so 51.7% of the documents are unique, which sounds reassuring until you notice they are the small ones. Those 1,678 single-use documents cover 1,678 series, which is 10.0% of the exchange. The other 15,083 series (90.0%) read a document written for a family.

Sharing is not evenly spread, and the pattern is the opposite of what you would guess. The categories with the most markets are not the worst offenders; the worst is a mid-sized category almost nobody writes about.

CategorySeriesDocumentsSeries per document
Mentions5022025.1
Exotics14114.0
Elections2,6163377.8
Financials1,3312036.6
Entertainment2,6574495.9
Sports4,3177875.5
Economics9231934.8
Commodities120264.6
Climate and Weather5361433.7
Politics2,5657373.5
Crypto306913.4
Health94392.4
Science and Technology3971892.1
Companies1601071.5

Mentions is the extreme at 25.1 series per document — 502 markets on 20 files. Mention markets are exactly the ones where the specific word matters most, and they have the thinnest document coverage on the exchange. At the other end, Companies runs 1.5 series per document, so a single-stock market usually does get its own text.

Every contract terms PDF we opened still has blanks

A shared template would be defensible if the template were complete — generic prose that is nonetheless true of every market using it. That is not what is in these files. We downloaded 30 contract terms PDFs, all returning HTTP 200, ran pdftotext over them and searched for unfilled merge tokens of the <event> variety.

All 30 contain at least one. Not 30 instances — 30 of 30 documents.

DocumentSeriesBlank tokensWords
SOCCERGAMEWIN2462454,703
SOCCERWHOSTAT222762,402
MENTION2061162,998
SOCCERSPREADS205662,234
SOCCERTOTALS188622,258
ACHIEVEMENTS181501,273
ENTITYOUTCOME122652,331
BASKETBALLGAMEWIN114802,348
RANKLIST10342889
NASCARGRIDMOVER11363,196
DEMOCRACYSTATUS114446
DRONESDOWN15382

Across the whole board the same shape shows in the API text: 4,749 series — 28.3% — publish at least one prohibition still containing a merge token, 10,355 instances across 41 distinct tokens, <event> alone accounting for 8,996. Against our 18 November reading of 4,715 series on a board of 16,626, that rate has not moved: 28.36% then, 28.33% now. Kalshi is not filling these in.

The product-certification twins are the same. MENTION.pdf in the certification path carries 133 blanks, PLAYERAGREEMENT.pdf 89. Google has indexed that second file under a title which is the placeholder text: “• <player's association> • <league> • <agreement>”.

One document, four different names

The 3,245 figure is itself an overcount of distinct documents, and we can prove it on the sample. We took the extracted text of all 30 PDFs and compared checksums. Thirty URLs returned 26 distinct document bodies.

Two clusters account for the difference. SOCCERADVANCE.pdf and SOCCERGAMEWIN.pdf are byte-for-byte the same text, between them the contract terms for 345 series. And one body is published under four names at once — ENTITYOUTCOME, NASCARCHALLENGE, NFLDIVISIONTOTALWINS and WCFASTESTGOAL — so a NASCAR challenge market, an NFL division win-total market and a World Cup fastest-goal market are all governed by one shared text of 2,331 words.

That is a 13% duplication rate on a 30-file sample. We have not extrapolated it to all 3,245 documents, because a sample weighted toward the most-shared files is not a random one, but it means the real count of distinct Kalshi contract terms documents is below 3,245 and the true series-per-document ratio is above 5.2.

“The binding document” needs a footnote

The best page on Kalshi's rule hierarchy is oddsshopper's August 2026 explainer, and it is genuinely good: a clean four-layer model from the market-card summary up through the expanded rules and the contract terms to the rulebook. It calls the PDF “the binding document” and sums the hierarchy up as “the contract terms govern, and the exchange interprets”.

Legally that is right. As instructions to a trader it stops one step short, and the example it picks is the sharpest case on the board. The file it links is MENTION.pdf — 206 series, 2,998 words, 116 unfilled tokens, <word> appearing 77 times. A reader who follows that link to settle a question about a specific phrase finds the phrase represented as a blank.

Our own earlier article made the stronger version of the same mistake, and it is worth stating plainly. The prohibited-traders census told readers the PDF “is authoritative where the API text has an unfilled placeholder” and called it “the per-series document”. Both halves are wrong: the document is usually not per-series, and it does not resolve the placeholder because it contains the same one. We are leaving that article in place and correcting it here rather than quietly editing it.

The accurate version of the advice is narrower. The contract terms PDF tells you the structure of the product — settlement mechanics, tick size, expiration timing, contingencies. It does not tell you which word, seat, team or threshold your own market is written on.

Where your market's real parameters live

Before calling any of this a failure we went looking for the filled text, because a template is only a problem if the specifics are published nowhere. They are published — as fields.

We walked 8,000 open markets through the markets endpoint. rules_primary, the field that would carry market-specific rule prose, was empty on 7,985 of them — 99.8%. Only 15 had anything in it, and none of those 15 contained a placeholder. The event object has no rules prose either. So there is genuinely no document and no API field that states your market's rule in filled sentences.

What is market-specific is the structured data, and it is good. A single mention series returns its own named settlement sources — ABC, Fox News, MSNBC, The Wall Street Journal, Semafor, The Information, CBS, The Washington Post — its own prohibition list and its own fee type.

What you want to knowWhere it actually isSpecific to your market?
Settlement mechanics, tick, expiry structurecontract_terms_url PDFNo — family template
Who decides the resultsettlement_sources on the seriesYes — named outlets
Who may not trade itadditional_prohibitionsYes, but 28.3% have blanks
What it costs to tradefee_type, fee_multiplierYes
The actual threshold or subjectMarket title, custom_strike, close_timeYes — but as data, not prose
Your market's rule in filled sentencesNowhere—

So the fair reading is not that Kalshi conceals anything. It publishes more machine-readable detail per market than any venue we have measured, including naming who settles each one. The gap is that the assembly job — template plus parameters — is left to the reader, and no page tells the reader that.

Checking your own market in one request

Two minutes of work gets you further than opening the PDF. Find your market's series ticker, then read the series object and see how many other markets share your document.

curl -s "https://api.elections.kalshi.com/trade-api/v2/series" \
  | python3 -c "
import json,sys,collections
S=json.load(sys.stdin)['series']
c=collections.Counter(s['contract_terms_url'] for s in S)
t='KXPELOSIMENTION'          # your series ticker
s=next(x for x in S if x['ticker']==t)
print('doc      :', s['contract_terms_url'])
print('shared by:', c[s['contract_terms_url']], 'series')
print('sources  :', [q['name'] for q in s['settlement_sources']])
"

No key and no bot check: the endpoint is public, and leaving the category parameter off returns the whole board. If “shared by” comes back as anything above 1, the PDF you are about to open was not written for your market. Read it for mechanics, then take the settlement sources and the strike from the fields.

My verdict

Kalshi publishes a genuinely unusual amount per market, and the two document fields are the weakest part of it. A field documented as “the URL to the current terms of the contract underlying the series” resolves, nine times out of ten, to a file covering other people's markets — up to 493 of them — with the distinguishing details left in angle brackets and at least one body shipped under four names.

None of that changes what you are bound by. The certified product template plus the rulebook plus your market's published parameters are the contract, and they are all available. What is missing is any single artefact that reads like your market's rules, which is precisely what every guide, ours included, has been telling you to go and find.

Trade off the fields, not the PDF. Use the document for mechanics and the series object for everything that identifies your market — and if a 25-series-per-document category like Mentions is where you are putting money, assume the text is describing a family and check the strike yourself.

Check the terms before you take the position

Settlement sources, prohibitions and fees are published per market on both venues. Reading them costs two minutes; finding out afterwards costs the position.

Visit Polymarket.com →

Affiliate link to Polymarket; we have no commercial relationship with Kalshi and earn nothing from it. 18+ only. Prediction markets are not available everywhere — check your own jurisdiction first.

Kalshi contract terms FAQ

Where do I find the contract terms for a Kalshi market?

From the Rules section on the market page, or directly from the contract_terms_url field on the market's series object in Kalshi's public API. Every one of the 16,761 series we read carries one. Expect a PDF named after a product family rather than your market.

Are Kalshi contract terms specific to each market?

Usually not. 16,761 series resolve to 3,245 distinct documents, and only 1,678 series — 10.0% — have a document used by no other market. The remaining 90.0% share. One file, ELECTION.pdf, is the contract terms for 493 series.

Why does my Kalshi contract terms PDF contain text in angle brackets?

Those are unfilled template merge tokens, and they are normal rather than rare. All 30 documents we sampled contained at least one. MENTION.pdf, which covers 206 markets, has 116 of them, including <word> 77 times. The token marks where a market-specific value would go if the document were generated per market.

Is the contract terms PDF the binding document?

It is the certified product template, and it governs the product's mechanics. It is not a complete statement of your market's rule, because it does not contain your market's values. Treat it as one of three parts alongside the KalshiEX rulebook and the market's own published fields.

If the PDF has a blank, what resolves it?

The series and market objects, not another document. The series carries named settlement_sources, its additional_prohibitions list and its fee type; the market carries the title, strike and close time. We checked whether filled prose exists anywhere in the API and found rules_primary empty on 7,985 of 8,000 open markets.

What is the difference between contract_url and contract_terms_url?

contract_terms_url points at the current terms under assets.kalshi.com/contract_terms/. contract_url points at the original CFTC product-certification filing under assets.kalshi.com/regulatory/product-certifications/. Both are templated: the certification copy of MENTION.pdf carries 133 blank tokens.

Which Kalshi category has the worst document coverage?

Mentions, at 25.1 series per document — 502 markets sharing 20 files. That is the category where the exact wording matters most. Companies is the best at 1.5 series per document, so single-stock markets usually do get their own text.

Does Kalshi publish the same document under more than one name?

Yes. Checksumming the extracted text of our 30 samples gave 26 distinct bodies. SOCCERADVANCE.pdf and SOCCERGAMEWIN.pdf are identical, and one 2,331-word body is served as ENTITYOUTCOME, NASCARCHALLENGE, NFLDIVISIONTOTALWINS and WCFASTESTGOAL.

Method and sources

All counts come from a single pass of Kalshi's public series endpoint on 20 November 2026, returning 16,761 series as complete objects with the category parameter omitted. Document analysis used 30 PDFs downloaded that day, text extracted with pdftotext -layout and compared by checksum. The 8,000-market rules_primary measurement walked the open-markets endpoint in eight pages. Figures are dated because this board grows: it was 16,626 series on 18 November.

Two caveats we will not paper over. The 30-PDF sample is deliberately weighted toward the most-shared documents plus a dozen single-series files, so its 13% duplication rate is not a board-wide estimate. And kalshi.com's own web pages returned HTTP 429 to this environment throughout, so the rulebook is cited from the Robinhood-hosted copy rather than from Kalshi directly.