UMA marketing: how the oracle behind prediction markets sells truth with 100,000 assertions, dispute data and a blog that teaches the category

How UMA markets its optimistic oracle for prediction markets with dispute data, 100,000 assertions, a blog of explainers, an AI truth bot and partners.

The Convrt team14 min read28 sources
100,000assertions made onchain
~1%of assertions disputed
50posts on the UMA blog
The short version

UMA is a decentralized oracle built by Risk Labs. Its whitepaper came out in December 2018, its optimistic oracle has resolved prediction markets since 2022, and in October 2025 it passed 100,000 assertions made onchain.

Its marketing sells one idea, a decentralized truth machine, and proves it with two numbers it repeats in almost every post: how many assertions it has settled and how few were disputed.

Around that sits a teaching engine: a blog of 50 posts that explains prediction markets, disputes, voting and P4, a public record of its rule changes with before and after data, and an AI truth bot that posts its answers on X.

UMA markets an oracle, the part of a prediction market that decides what really happened. This breakdown covers who UMA is, the story by the dates, how the optimistic oracle works, what builders use it for, 100,000 assertions, how rare disputes are, the blog, the explainers, P4 and the voting options, blind consensus, managed proposers, what the whitelist changed, the Optimistic Truth Bot, how accurate the bot is, 2024 in review, partners and chains, and what any prediction market brand can copy.

UMA at a glance
  1. 012018Whitepaper published, in December.
  2. 02100,000Assertions made onchain, Oct 2025.
  3. 03~1%Of assertions disputed, Mar 2026.
  4. 04177Whitelisted proposer addresses.
  5. 0550Posts on the UMA blog.
  6. 06132Videos on YouTube.
UMA at a glance, from the Risk Labs, UMA blog, Predict.fun and YouTube pages (read 19 Nov 2026).

“A decentralized truth machine.”UMA, home page (read Nov 2026)

From whitepaper to prediction markets

The Risk Labs timeline tells the story in short entries. It starts in December 2018 with UMA Whitepaper published. Risk Labs was founded in January 2019, and the timeline calls it A small, close-knit team with a dream to make market access ubiquitous. April 2020 brought the UMA protocol and token launch. The entry adds: This marked the beginning of our work in synthetic assets and the pioneering of the Optimistic Oracle (OO).

The oracle then proved itself in two places. The October 2021 entry says: We launched Across as a testament to the flexibility of UMA’s Optimistic Oracle technology in verifying any knowable data. The February 2022 entry reads: The Polymarket integration kicked off the use of the OO in prediction markets, demonstrating additional capabilities of the Optimistic Oracle. And in February 2025, the timeline notes: UMA and partners began researching the next-generation prediction market oracle.

Prediction markets were close to crypto from the start, and UMA says so. A post from October 2024 puts it this way: Since their initial shout out in the Ethereum Whitepaper, prediction markets have become a cornerstone in the DeFi space and one of the hottest topics in 2024.

The story by the dates
→ 01 Dec 2018 UMA whitepaper published
→ 02 Jan 2019 Risk Labs founded
→ 03 Apr 2020 Protocol and token launch
→ 04 Oct 2021 Across launched on the oracle
→ 05 Feb 2022 The Polymarket integration
→ 06 Feb 2025 Next-gen oracle research
The story by the dates, from the Risk Labs timeline (read 19 Nov 2026).
How a request is resolved
→ 01 Propose A statement with a bond
→ 02 Challenge A window of 2 hours to 2 days
→ 03 Accept True if no one disputes
→ 04 Dispute Match the proposer’s bond
→ 05 Vote Commit and reveal, 24 hours each
→ 06 Settle At 65% of staked UMA
How a request is resolved, from the UMA home page, FAQs and blog (read 19 Nov 2026).
What builders use the oracle for
  1. 01MarketsPrediction markets and sports.
  2. 02InsuranceInsurance protocols.
  3. 03CrosschainAcross settles intents with it.
  4. 04Real worldReal world assets.
  5. 05OOv2Requests answered by proposers.
  6. 06OOv3Assertions with their own data.
What builders use the oracle for, from the UMA home page and documentation (read 19 Nov 2026).
Assertions resolved, as reported
Sep 2024 22,000+
End of 2024 34,000+
Jun 2025 ~60,000
Sep 2025 90,000+
Oct 2025 100,000
Assertions resolved, as UMA reported them in each post (read 19 Nov 2026).

Disputes are rare, and UMA says so in every post

The second proof number is how few answers get challenged. UMA reports it in nearly every post, and the figure keeps falling. Only ~1.5% of proposals get disputed. The OO is already quite efficient, clocking in at a dispute rate of ~1.3%. In March 2026, the Predict.fun post put it at ~1% of all assertions get disputed.

It also separates real challenges from early or noisy ones. A May 2025 post says only about 0.4% of all assertions face substantive challenges on their accuracy. The Fact vs. Fiction post adds: Over 99% of all proposals are resolved without escalation. When disputes do arise, voters often reach >90% consensus.

Then it gives readers a picture they already know. Think of it like air travel. Thousands of flights land safely every day, but the few delays or cancellations grab headlines. UMA’s system resolves thousands of markets smoothly and leverages rare disputes to improve over time.

Dispute rate, as reported
Sep 2024 1.67%
2024 1.4%
Feb 2025 ~1.5%
Aug 2025 ~1.3%
Mar 2026 ~1%
Dispute rate in percent, as UMA reported it in each post (read 19 Nov 2026).
UMA blog posts per year
2024 27
2025 22
2026 1
UMA blog posts per year, from the blog archive (read 19 Nov 2026).

Explainers that answer search questions

The titles read like search queries: What is a Prediction Market?, What is a Prediction Market Dispute?, Why Prediction Markets Need Oracles and What are Prediction Market Clarifications? Each one answers a single question in plain words, then shows how UMA handles it.

The dispute post is a good example. Prediction market disputes introduce four main challenges into the DeFi space. It starts with the basic tension, Every prediction market creates winners and losers, and then lists fixes: Providing very clear, well-defined YES and NO market conditions. Defining credible sources for market resolution. Clarifying unclear or ambiguous market rules.

The clarifications post shows how the community fills the gaps. Clarifications are a community-led initiative from the UMA and Polymarket community. Less than 1.5% of proposals are disputed, and clarifications are only a fraction of those. The UMA DVM has never voted against a clarification.

Explainers that answer search questions
  1. 01MarketsWhat is a prediction market?
  2. 02DisputesFour challenges and three fixes.
  3. 03OraclesWhy prediction markets need them.
  4. 04ClarificationsLed by the community.
  5. 05P4Too early, and why it matters.
  6. 06DelphiThe history of oracles.
Explainers on the UMA blog that answer search questions (read 19 Nov 2026).
The voting options
  1. 01P1No.
  2. 02P2Yes.
  3. 03P3Unknown or 50/50.
  4. 04P4The proposal was made too early.
  5. 05WaitThe advice, in one word.
  6. 06Past tenseHow every proposal is written.
The voting options, from UMA Voting: What is P4 and Why Does it Matter?, 11 Feb 2025 (read 19 Nov 2026).
How a vote is counted
→ 01 Commit A hashed vote, 24 hours
→ 02 Reveal Vote and salt, 24 hours
→ 03 Mode Used if over 50% agree
→ 04 Median Used if no value passes 50%
→ 05 Rewards The middle 50% of voters
→ 06 Roll A new round below 65%
How a vote is counted, from The Power of Blind Consensus, the FAQs and Recent Improvements posts (read 19 Nov 2026).
Whitelisted proposer addresses
Sep 2025 37
Oct 2025 156
Nov 2025 177
Whitelisted proposer addresses, from Managed Proposers Update, 8 Nov 2025 (read 19 Nov 2026).
What the whitelist changed
  1. 0196%Of proposals came from whitelisted addresses.
  2. 0299.7%Their accuracy.
  3. 0385.8%Accuracy of everyone else.
  4. 0459%Fewer incorrect proposals.
  5. 0568%Fewer disputes.
  6. 065 at 95%Proposals and accuracy to join.
What the whitelist changed, from the Managed Proposers posts of 12 Aug and 8 Nov 2025 (read 19 Nov 2026).
How the Optimistic Truth Bot works
→ 01 Watch Reads Polymarket questions
→ 02 Router Picks the cheapest viable path
→ 03 Solvers Scripts in a 60-second sandbox
→ 04 Cite Sentences without a URL are cut
→ 05 Overseer Unsafe if the price says 85% the other way
→ 06 Post Shares the answer at @OOTruthBot
How the Optimistic Truth Bot works, from Inside UMA’s Optimistic Truth Bot, 30 Jul 2025 (read 19 Nov 2026).
Truth Bot accuracy, as reported
First prototype ~60%
Mention markets 72%
All markets ~78%
Excluding early or ambiguous ~95%
Sports and asset pricing 99.3%
Optimistic Truth Bot accuracy in percent, as reported on the UMA blog (read 19 Nov 2026).
2024 in review
  1. 0134,000Assertions processed.
  2. 0298.6%Resolved without disputes.
  3. 03~2,000Unique validator addresses.
  4. 0491%Reveal-to-commit ratio, up from 85%.
  5. 05868Critical votes secured.
  6. 06500Of those for Polymarket.
2024 in review, from the UMA blog post of 1 Jan 2025 (read 19 Nov 2026).
Partners, chains and new markets
  1. 012022The Polymarket integration.
  2. 02BaseIncluding Bookies, for sports.
  3. 03BlastStarting with Predict.Fun.
  4. 04BNB ChainPredict.fun, a multi-year deal.
  5. 05120,000+Users on Predict.fun.
  6. 06StoryProgrammable IP.
Partners, chains and new markets, from the Risk Labs, 2024 in review, Predict.fun and programmable IP posts (read 19 Nov 2026).

What any prediction market brand can copy

What any prediction market brand can copy
→ 01 Job Name the one thing you do
→ 02 Proof Repeat one number everywhere
→ 03 Explain One post per search question
→ 04 Change Publish the reasons and results
→ 05 AI Show the work, even weak scores
→ 06 Clips Cut every explainer into short form

UMA markets something most users never see: the step that decides who was right. It names that job in one line, a decentralized truth machine, and proves it with two numbers it repeats in every post, how many assertions it has settled and how few were disputed. Around that it teaches the category, publishes its rule changes with before and after data, and shows its AI experiments while they are still imperfect.

The lessons are simple. Name the one job you do. Pick a proof number and repeat it everywhere. Answer the questions people search for, one post each. When you change the rules, publish the reasons and the results. Show your work in public, even the weak scores. Then cut the clearest minute of every explainer, milestone and product update into clips and post them daily on TikTok, Instagram and YouTube Shorts from accounts built for short form. For how a prediction market newsletter explains the dispute process to readers, see our Polymarket The Oracle breakdown.

How we would run it for you Trading platform marketing, run for you

We run creator content and posting for trading platforms on accounts you own, built for your compliance sign-off, with every post tracked for views, link clicks and click-through rate.

Frequently asked questions

UMA is an optimistic oracle and dispute arbitration system that brings data onchain. It is built by Risk Labs, the team behind UMA and Across. Anyone can propose an answer with a bond, and the answer is accepted if no one disputes it during the challenge period. Disputes go to a vote of UMA tokenholders.

A proposer submits the outcome with a bond. Anyone can dispute it by matching the bond during a challenge window, which integrations set between 2 hours and 2 days. If it is disputed, tokenholders vote in a commit phase and a reveal phase of 24 hours each, and the result needs 65% of staked UMA to settle.

Rarely, by its own figures. UMA reported a 1.67% dispute rate in September 2024, 1.4% for 2024, about 1.5% in February 2025, about 1.3% in August 2025 and about 1% in March 2026. Its Fact vs. Fiction post says over 99% of all proposals are resolved without escalation.

Through teaching and proof. Its blog of 50 posts explains prediction markets, disputes, clarifications and P4 in plain words, most posts repeat a resolution number, and milestone posts mark scale such as 100,000 assertions. It also publishes the reasons and the results of changes like the proposer whitelist.

Name the one job you do, repeat one proof number everywhere, answer the questions users search for, publish the rules and the results when you change them, show your AI work in public, and cut the clearest minute of every explainer into short clips.

  1. UMA, home page, read 19 Nov 2026.
  2. Risk Labs Foundation, home page and timeline, read 19 Nov 2026.
  3. UMA Documentation, Welcome to UMA, read 19 Nov 2026.
  4. UMA YouTube channel, read 19 Nov 2026.
  5. UMA blog, post archive by date, read 19 Nov 2026.
  6. UMA blog, 100,000 Assertions and Counting, 7 Oct 2025, read 19 Nov 2026.
  7. UMA blog, UMA Powers Market Resolution for Predict.fun on BNB Chain, 10 Mar 2026, read 19 Nov 2026.
  8. UMA blog, Improving Oracle Efficiency with Managed Proposers, 12 Aug 2025, read 19 Nov 2026.
  9. UMA blog, Managed Proposers Update, 8 Nov 2025, read 19 Nov 2026.
  10. UMA blog, 2024 in Review: Truth, Trust, and Innovation at UMA Protocol, 1 Jan 2025, read 19 Nov 2026.
  11. UMA blog, Fact vs. Fiction: The Critic’s Guide to UMA, 9 Sep 2025, read 19 Nov 2026.
  12. UMA blog, The Power of Blind Consensus, 10 Jun 2025, read 19 Nov 2026.
  13. UMA blog, Inside UMA’s Optimistic Truth Bot, 30 Jul 2025, read 19 Nov 2026.
  14. UMA blog, AI Is Helping Us Find the Truth, 30 Jun 2025, read 19 Nov 2026.
  15. UMA blog, UMA’s AI Experiment, 3 Jun 2025, read 19 Nov 2026.
  16. UMA blog, What are Prediction Market Clarifications?, 4 Mar 2025, read 19 Nov 2026.
  17. UMA blog, UMA Voting: What is P4 and Why Does it Matter?, 11 Feb 2025, read 19 Nov 2026.
  18. UMA blog, UMIP 183: Streamlining Web3 Prediction Markets, 27 Jan 2025, read 19 Nov 2026.
  19. UMA blog, What is a Prediction Market Dispute?, 6 Sep 2024, read 19 Nov 2026.
  20. UMA blog, FAQs: Everything You Need to Know About UMA, 19 May 2025, read 19 Nov 2026.
  21. UMA blog, Case Study: How UMA Secures Across Protocol, 29 May 2025, read 19 Nov 2026.
  22. UMA blog, What is UMA’s Optimistic Oracle?, 18 Feb 2025, read 19 Nov 2026.
  23. UMA blog, What is a Prediction Market?, 16 Aug 2024, read 19 Nov 2026.
  24. UMA blog, Why Prediction Markets Need Oracles, 7 Oct 2024, read 19 Nov 2026.
  25. UMA blog, The Economics of Truth, 28 Apr 2025, read 19 Nov 2026.
  26. UMA blog, From Delphi to DeFi: The Evolution of Blockchain Oracles, 17 Mar 2025, read 19 Nov 2026.
  27. UMA blog, The Future of IP is Programmable, 6 Aug 2025, read 19 Nov 2026.
  28. UMA blog, Recent Improvements to UMA’s Optimistic Oracle, 8 May 2025, read 19 Nov 2026.

More breakdowns

All breakdowns →

Want distribution done properly?

Apply and we'll map your existing content to real reach on accounts you own.

Pinch to zoom · swipe down to close

Apply to work with us.

Five quick questions, then pick a time.