Sixty percent. That is the number circulating this week. A merger, we are told, has a 60% chance of happening because Kalshi's prediction market says so. The number has moved from a little-known event contract to a mainstream financial headline without a single question about what sits behind it. Let's look at the data. Actually, let's look at what is missing.
The claim, as parsed from the source material, contains no contract expiration date. It contains no volume figure. No open interest. No bid-ask spread. No settlement definition. The source report itself concedes that only three data points were extracted from the original article, and none of them answer the questions that matter. That absence is not a minor oversight. It is the story.
A prediction market price is a capital-weighted opinion, not a statistical probability. Treating a single price as a verified forecast is the crypto equivalent of reading a token's market cap and concluding the protocol is solvent. Rigour over rumour.
Data Integrity Check: What We Are Not Being Told
I have spent the past eight years building standardized frameworks to separate signal from noise. The first thing I do when a number arrives without a protocol layer is run it through a simple checklist. There are seven fields I need before any prediction market percentage can be treated as evidence. Contract ID. Expiration timestamp. Settlement rule. Last trade price. Volume over the trailing 24 hours. Open interest. Best bid and ask. The source material provides exactly one of those: last trade price. That is not data. It is a data fragment.
A data fragment can be true. It can also be meaningless. The difference is not the number; it is the context around the number. A 60% probability from a contract with 100 open contracts and a three-cent spread is not the same as a 60% probability from a contract with 500,000 open contracts and a one-cent spread. Both numbers are 60. Only one tells you something about the world.
This is why the 'Data Integrity Check' is the first section of every major report I write. It forces the writer to admit what is unknown before claiming what is known. In this case, the unknown far outweighs the known.
What Kalshi Actually Is
Kalshi is not a crypto casino. It is a CFTC-regulated exchange for event contracts. It operates under a Designated Contract Market license, which means every contract listed has been cleared for trading. A binary contract, such as 'Will this merger close by December 31?', trades from zero to 100 cents. If the event occurs, each yes contract pays $1. If not, it pays $0. The market price, when read as a probability, is simply the last place where marginal buyers and sellers stopped fighting.
That structure is clean. It is auditable. It is also easily misunderstood. The price of a binary contract is a consensus of the bid-ask spread, but only among the traders who are actually present. If a venue requires KYC, deposits via bank transfer, and has a clunky interface, it is not a representative sample of the broader market. It is a sample of people who are willing to complete KYC. That is a subtle selection bias, and it matters.
Kalshi's regulatory license is a genuine moat. Polymarket cannot legally offer the same products to U.S. traders. PredictIt is limited to academic and political event contracts. Kalshi has the CFTC's blessing, a bank-compatible settlement layer, and a growing menu of event contracts. That is an asset. But a moat is not a market. A license is not liquidity.
I have spent fifteen years in this industry. In 2017, I audited fifteen ERC20 whitepapers and flagged eight projects whose distribution models were mathematically broken. The market did not care for the first month. Then the distribution models broke the prices. I learned that the same data discipline that applies to tokenomics applies to event contracts. A price without a protocol layer is just a number. You cannot verify it; you can only quote it.
The 60% figure fails the first test. Let's break down why.
Step 1: Define the Contract
The phrase 'merger probability' is a category error unless the exact contract is named. A Kalshi event contract has a precise definition. 'Will Company A announce a merger by the end of Q3?' is a different instrument than 'Will Company A close its pending merger with Company B within 90 days?' The market can price one at 60 and the other at 20. If a headline quotes a single '60%' without specifying the contract, the number has already lost its informational content.
This is not semantics. During my ICO audit work, I saw the same problem in whitepapers: projects would present 'token supply' without specifying whether it was total supply at genesis, circulating supply after lockups, or inflation-adjusted supply. The resulting valuations were nonsense. Prediction market probabilities are no different. The settlement definition is the denominator of every calculation. If you don't know the settlement rule, you don't know the probability.
Step 2: Demand Time-to-Expiry
A probability is a function of time. A 60% chance of a merger happening within the next month is a completely different risk from a 60% chance of the same merger happening in the next 18 months. The market prices both, but they trade at different levels. The report I was asked to audit does not give an expiration. That means I cannot evaluate whether the price is fair, stale, or simply misquoted.
In 2020, I built an Excel model to track Compound Finance's yield rates across fifty liquidity pools. I found a 15% arbitrage between ETH and DAI pairs. The opportunity existed only because I had standardized the time component. Every position had a start date, a maturity date, and a compounding rule. Without time, the yield figure is meaningless. The same logic applies to Kalshi. A contract with a shorter time-to-expiry will trade at a lower price than one with a longer time-to-expiry, all else equal. A headline that does not include the expiration date is not a data point. It is a click.
There is another problem with time: a probability can be 'true' at one moment and false an hour later. The source material does not even include a publication timestamp. In this industry, an un-dated price is almost worthless. You cannot audit a 60% reading at 9:00 AM if the merger announcement was released at 2:00 PM. The price would have moved. The headline would still say 60.
Step 3: Check the Order Book Health
This is where the 60% number starts to become dangerous. Prediction market contract prices are set by actual buys and sells. But on a lightly traded contract, a single trader can move the last price by ten points. A $2,000 market order can transform '55%' into '65%' in less than a second. That price then becomes a screenshot, then a tweet, then a news article.
I have written extensively about this liquidity trap. The reliable threshold I use is blunt: if a prediction contract has fewer than 10,000 contracts of open interest and fewer than 2,000 contracts in daily volume, it is a niche instrument, not a market. At that level, the price is a quote, not a consensus. It tells you where one marginal participant was willing to trade. It does not tell you anything about the aggregate probability of a merger.
The source report's own first paragraph admits that volume and open interest were not extracted. So we have no way to verify whether 60% is a robust consensus or a stray print. The absence of these metrics is itself a red flag. In my crisis work during the Celsius collapse, I monitored 200+ smart contract wallets for sudden outflows. The signal that mattered was not a single wallet leaving. It was the rate of change across a basket of wallets. A single data point is noise; a distribution is information. The 60% figure is a single data point.
Step 4: Push on the Spread
The bid-ask spread is the closest thing a prediction market has to a confidence interval. A liquid contract trades at a one-cent spread. The market is saying: we agree on the price within a narrow range. A thin contract trades at a five-cent spread or wider. The market is saying: we are not sure where the price is, and we are charging a premium for the uncertainty.
If the last trade on Kalshi printed at 60 cents, but the best bid is 55 cents and the best offer is 64 cents, the reported '60%' is a midpoint that has not actually traded at that level. The headline becomes a rounding error. This is not a niche nuance; it is the difference between a market-determined probability and an accountant's average. I teach every analyst who works with me to pull the level 2 book before quoting any binary contract. Level 2 data is messy. It is also honest.
Step 5: Cross-Check Against Independent Venues
Kalshi is regulated, but it is not the only prediction market. Polymarket runs on Polygon, with transparent order books and on-chain settlement. PredictIt has been quoting political and economic event contracts for years. If a merger is a major event, these markets should tell similar stories. They will not be identical. They should not be. But a divergence wider than ten percentage points is a clear warning that one of the venues is pricing a different contract, a different timeline, or a materially thinner book.
When I led the AI clustering project at Dune Analytics, we standardized wallet classification across institutional and retail entities. The raw data was messy; the output needed to be comparable. The same is true for prediction market prices. You cannot compare a Kalshi price to a Polymarket price without normalizing the contract definition. If you do, you are fooling yourself. In this case, the headline '60% merger probability' has not been cross-validated. We are being asked to accept one venue's price as a fact. I refuse to do that. Check the chain, not the hype.
Step 6: Understand Who Is Trading
The Kalshi order book is not a random sample of public opinion. It is a group of people who have passed KYC, linked a bank account, and chosen to trade event contracts on a CFTC-regulated exchange. That selection effect matters. Kalshi's regulatory status is a moat against Polymarket, but it is also a toll booth. It filters out the retail traders who might otherwise provide liquidity. The result is a book that is often institutional and occasionally desolate.
This is where my 'KYC is theater' argument gets more nuance. In crypto, project KYC is often just a screenshot of a passport; the actual compliance layer is laughable. Kalshi's KYC is not theater. It is a real barrier. But the barrier does not make the price more correct. It just makes the book thinner. A venue with high compliance costs and low participation can produce a quoted probability that is less informative than a no-KYC venue with a deep, active book. Regulated does not mean robust.
The cost of that compliance falls on the users who choose the regulated route. The very traders who are trying to do the right thing become the only ones required to prove who they are. That is a tax on honesty. It is exactly the kind of structural inefficiency that I have been documenting since my first ICO audit. The market's first line of defense is not regulation. It is transparency.
Step 7: Run a Verification Score
Let me make this concrete. I use a simple verification score for any prediction market signal. It is not a model; it is a filter. The formula is:
Verified Probability = Midpoint Price min(1, OI / 10,000) (1 - max(0, Spread - 0.01) * 5)
Midpoint price is the average of the best bid and best ask. OI is open interest. Spread is the difference between the best ask and best bid, expressed as a decimal.
At 10,000 open interest and a one-cent spread, the score equals the midpoint. At 2,500 open interest and a six-cent spread, the score is 0.25 0.75 = 0.1875, an 18.75% probability, not a 60% one. At 500 open interest and a twelve-cent spread, the score is 0.05 0.45 = 0.0225, a 2.25% probability. That is the difference between rigour and rumour.
The formula is deliberately conservative. It penalizes thin books and wide markets. It refuses to allow a single price to be quoted as a probability. I have been using a version of this filter since my 2020 yield aggregation work. It would have saved a lot of bad trades.
Step 8: Check the Settlement Source
The market can be perfect and the settlement can still ruin the signal. A prediction contract is only as good as its settlement rule. Kalshi picks an authoritative source for each event, such as a news service, a court filing, or a regulatory database. If that source is ambiguous, the contract becomes a game of lawyers, not a forecast.
Most people ignore settlement definitions because they assume an event either happens or it doesn't. A merger is not that binary. Announcement dates get pushed. Regulatory approvals get extended. Governments block deals at the last hour. If the contract uses 'public announcement of a signed definitive agreement' as its trigger, a leak from an anonymous source is not enough. If it uses 'completion of the merger', then a signed agreement is not enough. The difference between 60 and 20 could be one missed word.
I learned this lesson in 2021 while standardizing NFT rarity scores. I discovered that 'background' attributes had a 20% higher correlation with long-term price stability than 'fur'. The reason? The field called 'fur' was inconsistently named across the metadata. Some projects kept a JSON string; others used an integer. The label was unreliable. Prediction contract settlement is the same. The trigger phrase is the metadata. If the metadata is ambiguous, the probability is a guess wearing a data costume.
A Liquidity Example
Let's make the verification score concrete. Suppose a news outlet quotes Kalshi at 60%. I pull the order book. The last trade printed at 60 cents. The best bid is 58 cents. The best ask is 62 cents. Open interest is 2,000 contracts. The 24-hour volume is 500 contracts. The midpoint is 60 cents, but the spread is four cents.
My verification formula gives:
60 (2,000 / 10,000) (1 - (0.04 - 0.01) 5) = 60 0.2 * 0.85 = 10.2%
A 10.2% verified probability, not 60%. The market is not saying what the headline thinks it is saying. It is saying that a small group of traders, none of whom are under pressure to take the other side, produced a price in a thin book.
That is the gap between a quote and a signal. This is why I insist on a standardized methodology. The number '60%' is not false; it is incomplete. But an incomplete number is not a neutral failure. It is a source of false confidence.
The Dune Query That Would Solve This
In my current role at Dune Analytics, I would handle this signal by creating a standardized query. The query would take a contract address, a timestamp, and a liquidity threshold. It would then pull the midpoint price, the open interest, and the spread. I would output a single column called verified_probability. That is the number I would trust. The query takes less than a minute to write. The reason no one writes it is not technical. It is commercial. A verified probability of 10% does not generate clicks. A clean 60% does.
I standardized a version of this for enterprise clients, reducing query time by 40%. The lesson was simple: the problem was not the data. It was the willingness to accept a raw price as an insight. If you want a probability, you must earn it. Earn it with volume thresholds. Earn it with spread filters. Earn it with settlement checks.
The Business Model Behind the Signal
There is another layer to this story. Kalshi earns fees on every contract. A busy event contract with a tight spread is a profitable product. The media habit of quoting a single '60%' is a marketing gift. A prediction market that gets quoted by mainstream finance becomes a prediction market that gets more trading volume. That is not a conspiracy. It is an incentive structure.
The source report's own dimensional analysis flagged Kalshi's business model as medium relevance and user growth as low relevance. I would argue that the business model is exactly why the 60% number is circulating. There is no free probability. Every quoted number is a product in disguise. This is not to say Kalshi is manipulating its price. It is to say that the ecosystem around prediction markets has an inherent interest in making the numbers easy to quote. Easy-to-quote numbers are the ones that get attention. Attention drives order flow. Order flow drives fees.
I have seen this dynamic in crypto countless times. A project with an attractive APR is not a project with a sustainable yield. It is a project with a marketing budget. The same logic applies here. A 60% headline is not a forecast. It is a customer acquisition channel.
There is also a reflexivity problem. The more the same percentage is repeated, the more the market trading on that percentage begins to shape the actual company's decision. A CEO reading 'the market is pricing 60% merger odds' may interpret that as pressure. An activist investor may use the quote as evidence of market support. The prediction market stops being a thermometer and becomes a thermostat. This is precisely why a skinny contract with a single quoted price can influence a real outcome. The signal does not have to be accurate. It has to be repeated.
The Contrarian Angle: 60% Is Too Neat
Now for the contrarian angle. The market's real signal is not the '60% probability' itself. It is the absence of confirmatory data structure.
In fact, the contrarian view here is that 60% is probably too neat. Prediction market probabilities that are quoted as neat round numbers are often the result of a mid-point calculation or a broadcast error, not a true market consensus. A real order book produces prices like 61.5, 58.25, or 63 cents. A perfect '60%' suggests someone has simplified the data. In my 2017 ICO work, the most suspicious token allocations were always the ones that came in neat percentages like 20/30/50. The messy ones, with vesting cliffs, lockups, and conversion formulas, were usually more honest. The same heuristic applies to prediction markets. Clean numbers are the first sign of data laundering.
I have seen this exact pattern in NFT markets. In 2021, I analyzed 10,000 Bored Ape transactions to create a standardized rarity score. The floor price, the cheapest listed NFT, was constantly quoted as the market's valuation. But the floor price was often a single outlier listing, not a trade. The real signal was in the distribution of sales, not the lowest ask. The same is true for Kalshi. The '60%' is the floor price of a prediction market. It is the cheapest marginal opinion, not the weighted distribution of opinions.
The deeper problem is correlation and causation. Let's suppose the Kalshi price is exactly 60 cents and the merger does happen. We might say the market predicted it. But that is confusing a market-clearing price with a causal forecast. A 60-cent price does not make a merger more likely; it merely reflects the highest bid and lowest ask at a particular moment. When the merger news hits, the contract will jump to 95 cents or fail to $1. But that jump is not a reward for correct prediction. It is a re-pricing of settlement risk. The price did not 'know' anything. It was just standing in the queue.
This is the blind spot in every event-contract headline. Prediction markets are often celebrated as 'Wisdom of the Crowd' devices. But the crowd is not always wise. The crowd is frequently indifferent. On a low-volume contract, the crowd is one person with a fat wallet and an opinion. The 60% figure might be nothing more than a trader's ability to post a limit order on one side of the book. That is not wisdom. It is market structure.
I saw the same dynamic in the 2022 Celsius collapse. We had a $12 million drain from Lido's stETH pool 48 hours before broader panic. The signal was not a dramatic price crash; it was a flow anomaly. The market still looked calm. A single massive outflow was the tell. If we had waited for the consensus price to move, we would have been too late. The lesson: the most important data is often the data that is absent. In this case, the absence of volume and spread data is that same kind of tell. It tells me that the 60% number is more marketing than measurement.
Crisis Protocol: Pre-Defined Triggers
Because I run every major market report through a crisis protocol, I am going to give you the exact triggers I use when evaluating any Kalshi-style probability signal. These are not recommendations. They are rules.
First, if the source cannot provide a contract identifier, an expiration date, and a settlement definition within two clicks, treat the signal as unverified. Do not trade on it.
Second, if open interest is below 5,000 contracts, assume the price is fragile. A single market order can create a false breakout. If open interest is below 1,000 contracts, ignore the price entirely. You are looking at a random walk.
Third, if the bid-ask spread is wider than five cents, the market is admitting its own uncertainty. Do not use the last traded price as your probability. Use the midpoint, and then apply a haircut.
Fourth, if Kalshi and Polymarket diverge by more than 15 percentage points, do not pick one. Step away. The contracts are probably not identical, or one of the venue's order books is compromised. There is no arbitrage signal here; there is only noise.
Fifth, and this is the one I care about most: if the number appears in a headline without any of the above data points, do not repeat it. Repetition is how false precision enters the market. Every time a news outlet writes 'the market is pricing 60%', it adds a layer of credibility that the market never earned. You are not reporting on data. You are manufacturing it.
I have made my living for fifteen years by standardizing messy information into auditable frameworks. I do not care about being entertaining. I care about being correct. The 60% merger probability might be correct. But 'might be' is not a basis for capital allocation. Yield follows logic, not luck.
Takeaway
Next week, when someone quotes a prediction-market probability, ask four questions: Which venue? Which contract? Which expiration? Which order book? If the source cannot answer, ignore it.
The on-chain record, such as Polymarket's settlement receipts, Dune's indexed volume tables, or even the CFTC's daily report if you can get it, is the only place where a probability can be properly audited. The number is not the signal. The structure is. Check the chain, not the hype. And if the chain is too thin to inspect, treat the percentage as what it is: a headline. That is not a forecast. It is a fragment. Rigour over rumour.