The prior chapter explained why chains matter for transparency and censorship resistance. This one walks the birth of a market: how a human question becomes tradable outcome tokens, who controls parameters, and where ambiguity is locked in before the first trade.
Retail traders rarely deploy contracts—but you always inherit creation choices. Creation is where product shape meets oracle adapter: a clean binary on CPI is a different machine than a four-outcome nomination tree, even on the same chain. fee tiers, end timestamps, outcome cardinality, collateral token, and which oracle adapter was wired at deploy. Retroactive “clarifications” are dispute bait.
From question to open market
Creation is a pipeline, not a single click. Someone drafts plain-language rules and picks a product shape—binary, categorical, conditional, or bundled. Parameters are set: outcomes, end time, fee basis points, resolver address, and often an off-chain URI for full resolution text. A contract call registers the market and mints the outcome-token structure. Liquidity may be seeded by a pool deposit, protocol subsidy, or an empty curve that prints a meaningless fifty-cent mid until real capital arrives. Trading opens; prices discover. After the real world resolves, the oracle path fires and tokens redeem.
Kalshi-style venues compress this: compliance and operations approve templates—you trade curated listings instead of deploying. Polymarket-style hybrids list quickly with off-chain rules linked on-chain. Augur-style permissionless creation lets any wallet with gas publish—maximizing freedom and tail risk on do-it-yourself wording.
Who can create?
Permission models shape your risk before you read a single rule. Exchange curation trades flexibility for cleaner macros and fewer traps. Protocol allowlists sit in the middle. Permissionless listing maximizes long-tail experiments and invalid rates when creators paraphrase headlines. DAO votes add delay and political listings.
Legal wrappers constrain who may use a front-end—not always what exists on-chain. A market can persist while your app geo-blocks. That gap is why “the market exists” is not the same as “I can trade it legally or practically.”
Parameters that move your economics
Outcome count determines whether you are in a true binary or a mutually exclusive set where only one leg pays. End time is a Unix timestamp on-chain—compare it to the event you think you are forecasting, including reporting delays. Collateral choice imports bridge and depeg risk. Fee percentages split protocol and liquidity providers. The resolver address points to optimistic oracle, stake-weighted reporting, multisig, or centralized operations—different tail shapes for the same headline.
Resolution source text is often off-chain. That is where disputes start.
Worked example: clean macro versus sloppy prop
Creator lists: “Will CPI year-over-year print at or above 3.0% for the March release?” Outcomes are YES and NO. End time covers the Bureau of Labor Statistics publish window plus a short buffer. Rules cite the official series. An optimistic oracle adapter is wired at deploy. Ambiguity is low; dispute tail exists but is manageable with size caps.
Contrast: “Will Candidate X dominate the debate?” Dominance is undefined, polls disagree, and bond fights cluster on similar wordings. On permissionless venues, invalid or refund paths are first-class—not a rare glitch.
Categorical and conditional markets
A four-way nomination market deploys four outcome tokens. Economically, the outcomes are perfectly correlated: exactly one should win. Traders sometimes treat legs like independent binaries and build hedges that do not net the way they expect. One oracle assertion must map cleanly to one winner; mid-race dropouts need rules for reopening or voiding.
Conditional markets multiply failure modes. If the condition never triggers, refund rules matter. Bundles tie multiple contracts; one ambiguous leg poisons the basket. Unless each leg is as crisp as a macro, skip do-it-yourself bundles.
Liquidity at birth and gas
An empty pool shows a fake mid until someone deposits. A subsidized launch can tighten spreads but attract mercenary liquidity that leaves when emissions end. Where log-scoring or constant-product parameters apply, read reserves—do not trust a screenshot.
Deploy gas is paid by creators but shows up as thin long-tail markets traders never touch. Mint and redeem gas is paid by you. A three-cent edge on two hundred dollars notional can turn negative after swap and redeem fees on a small layer-two ticket.
Before your first trade
Identify the permission model. Open the rule URI and match it to the UI; note hashed metadata when available. Count outcomes and map hedges to the product shape. Score ambiguity honestly—high scores mean quarter-Kelly or skip. Compare end time to the event clock. Note resolver type and creator dispute history. Check seed total value locked and early volume. Cross-venue comparison only after resolution appendices match.
Contract quality is a separate forecast from your world view. A brilliant macro call through a sloppy question still loses to invalid or bond wars.
Fees, subsidies, and who pays for day-one depth
Creators sometimes pay deploy gas themselves; protocols sometimes rebate listing fees to attract controversial questions. Traders still pay mint, swap, and redeem gas on many designs. A market launched with a two-week emissions boost can show tight spreads on day one and eight-cent spreads on day twenty when mercenaries leave. Creation parameters are not static—governance can change fee tiers or oracle adapters later, which is another form of rule risk worth watching in announcements.
Frozen at deploy
Once live, you generally cannot change outcome count or collateral token without a new market. End time extensions are rare and disputed. “Clarifications” posted only on social media do not update the contract. Treat the deploy snapshot as the entire contract you own until redeem—everything else is commentary.
Narrative: two creators, two tails
Creator A deploys a Fed funds binary with templated rules, UMA adapter, and fifty thousand dollars of seed liquidity. Creator B deploys “Will the bill pass?” without defining which chamber or vote type. Trader one earns a few cents on A after fees. Trader two on B faces invalid or a bond fight that eats the edge. Same chain, same gas token, opposite contract quality. Creation is where that split is born.
Common mistakes
Trading day-one illiquid pools without simulating the curve. Assuming UI text equals the contract. Ignoring outcome count on categoricals. Arbitraging a new list against a regulated venue without a rule diff. Skipping creator track record when dispute history is data.
Governance after deploy
Token votes can change fee switches, oracle adapters, or fee recipients. You did not sign a static contract forever—you signed the version live at entry plus governance risk. Watch announcements the same way you watch polls.
Synthesis
You will not deploy most markets you trade, but you always live inside someone else’s deploy choices. Treat creation metadata as part of the contract: outcomes, clock, collateral, resolver, rules URI, seed liquidity. If any field fails scrutiny, the mid is decoration.
What comes next
Creation freezes outcomes, clocks, collateral, and oracle hooks—ambiguity at birth drives everything in settlement and arbitration later.
Why creation literacy beats chain literacy
Knowing which chain a market uses matters less than knowing what was frozen at deploy. Bridges and gas change; outcome count and resolver address do not. Creation literacy is the skill of reading that frozen snapshot as carefully as you read a poll crosstab.
Practice note
Pick one market you traded recently and list five deploy fields: outcomes, end, collateral, resolver, rules link. If you cannot fill all five from memory, you sized without creation literacy.
Reader takeaway
Before the first fill, read the deploy snapshot: permission model, outcomes, end time, resolver, rules URI, seed liquidity. Ambiguity at birth is dispute and invalid risk at settlement. Contract quality is as important as your event model. The oracle chapter next explains how those deploy-time adapters turn headlines into redeemable state.
Next: Oracles: Bringing Real-World Data to Smart Contracts