How over 10,000 prediction markets become stock analysis.
Every day we screen over 10,000 prediction markets. Ten make it into the report, and each gets its own market analysis with a path to the stock. This page explains how that selection happens – and how to read the result.
What a prediction market is
On a prediction market, participants trade contracts on the outcome of real-world events: rate decisions, inflation data, commodity prices, elections, geopolitical developments. A contract pays out a fixed amount if the event happens, otherwise nothing.
That leads to the decisive property: the price is the probability. If a contract trades at 64 cents, traders are pricing the event at roughly 64 percent. That number isn't an opinion or a survey response, but a price at which people actually trade.
That's exactly the difference from the information sources you'd otherwise use. A headline and an analyst's comment are quickly written. A quote only exists once someone trades at that price.
Surveys and analyst reports appear periodically and with a lag. A prediction quote is a continuously tradable consensus that moves with every new piece of information. You're not seeing what someone thought was likely three days ago, but what the market is pricing in right now.
Which markets are relevant for the stock market
Not every prediction market has anything to do with stocks. Kalshi also lists markets on sports results, entertainment topics, and curiosities. For us, one question decides: does the event move stock prices or entire sectors?
Monetary policy
Rate decisions by the Federal Reserve, the ECB, and other central banks, meeting dates, expected step sizes. The most direct channel into valuations, financials, and rate-sensitive industries.
Inflation and the labor market
CPI readings, core versus headline, the unemployment rate. Determine how much room the central bank has.
Commodities
Oil and agricultural prices, price thresholds, supply disruptions. Act directly on energy, transport, chemicals, and consumer goods.
Geopolitics
Trade routes, sanctions, conflict developments, nuclear agreements. Hit defense, energy, and logistics at once, often in opposite directions.
Fiscal policy and regulation
Tax proposals, legislative proposals, government intervention. Change the valuation assumptions of entire industries.
Elections and shifts in power
To the extent they decide majorities and therefore policy paths — not as political coverage, but as a leading indicator for regulation and fiscal policy.
Technology and AI
Model rankings, app downloads, usage shares. Hit the largest companies on the market directly.
Companies and deals
IPOs, quarterly metrics, deliveries, customer numbers. The shortest path from a contract to a single stock.
Crypto
Bitcoin, Ether, XRP, and others. Linked to the stock market through crypto ETPs and listed crypto companies.
We exclude sports, entertainment, and curiosities. A market on who wins a TV show may have high volume – for your stock analysis, it's worthless.
How we filter: from over 10,000 to 10
The work isn't in collecting. It's in leaving things out. On October 29, for example: 13,703 open markets, 10,228 past the objective screen, grouped into 3,550 events, 257 movement candidates, 10 in the report.
We source market data directly from Kalshi under official license. No scrapes, no detours, no delay through intermediaries.
Every market is assigned to a topic cluster or excluded. The assignment follows a fixed rulebook based on contract characteristics and categories — not a language model's judgment.
We compare the current implied probability against where it stood 24 hours and seven days ago. Only what has moved more than the market's own bid-ask noise gets selected. A market that hasn't moved contains no new information.
A single event often has several contracts on Kalshi with different thresholds. We score them together as one event and show the contract with the strongest move. In the report, "rungs" shows how many contracts belong to the event.
The remaining events are weighted. At most two per topic channel and one per underlying, so a single topic doesn't fill the report. Up to three slots can go to markets on leading business stories the movement missed. What's left are the day's ten moves that matter to stock investors.
Why rule-based, not AI-driven
A rulebook has a property a language model can't offer: the same data produces the same selection. Every selection decision can be traced back to a rule and verified. There are no invented connections, no day-to-day mood, no dependence on exactly how a question happened to be phrased.
For analysis data, reproducibility isn't a technical detail. It's the precondition for trusting the numbers week after week.
How we weigh things: the scoring
A large percentage move is not automatically a strong signal. Our score evaluates every move on five factors:
Size of the move
How clearly the implied probability has changed – over 24 hours and over seven days.
Relevance for stocks
How directly the event carries over to listed companies. A rate decision acts more broadly and more clearly than a local personnel story.
Liquidity
How much is traded. A low-volume market can swing sharply on just a few contracts. Such swings carry less information and are weighted lower.
Time to settlement
Contracts nearing expiry move mechanically hard, because they run toward 0 or 100 percent. These moves look dramatic but rarely contain new expectation.
Recent coverage
Whether there is news on the event right now that makes an analysis worthwhile.
The score decides which markets go into the report. The biggest move doesn't win – the most meaningful one does. The five factors are the plain-language summary: behind them is a rulebook with many thresholds, weightings, and exceptions that took us half a year of development and testing before the selection held up reliably, day after day.
The market analysis
This is where the real value is created. A quote alone tells you nothing about your portfolio. On top of that, contract titles are often not what they seem. That's why each of the ten markets gets its own entry in five steps:
According to Kalshi's rules, including the cases that might surprise readers.
Dated research, every statement with a source.
A clear verdict on whether the news supports the traded probability.
The strongest argument against the market's expectation.
Link by link, how the outcome reaches a stock, ETF, or ETP, and which link is most likely to break.
Will the US average diesel price be above $6.30 a gallon on October 31?
Only motoring club AAA's single daily reading on October 31 counts, not a monthly average or the government's weekly survey. Exactly $6.30 resolves "No".
AAA's average stood at $6.35 on October 28, about five cents above the threshold. Wholesale diesel has fallen 59.5 cents from its October 22 high.
If stations pass on just one tenth of that wholesale drop by October 31, the average lands at about $6.29, below the threshold.
Diesel above $6.30 → fuel costs stay high into October's inflation data → rate expectations → valuations of US freight and transport companies. The weakest link: a few cents on one day barely move monthly inflation.
Why we always show the other side
A news article usually tells you one side, often the more dramatic one. Anyone who only sees the market's expectation misses what speaks against it. That's why every market comes with the other side – and for events with an open outcome, the consequences of both cases.
Where AI contributes — and where it doesn't
Selecting markets and weighing them is rule-based. The report's text – summary, research for each market, and equity transmission – is drafted with AI assistance; the report says so. We consider this distinction important enough to state openly rather than bury in fine print.
How to read the report in five minutes
The report is built so that you get more concrete as you move down. You read it online in your browser. If you want to keep it, download it as a PDF. If you're short on time, the first page is enough.
Movers Table: how to read the columns
| Column | Meaning |
|---|---|
| Prob | Current implied probability of "Yes" |
| 24h / 7d shift | Change over 24 hours / seven days, in percentage points (pp) |
| Volume | Contracts traded in the last 24 hours and open interest (OI), for the whole event – in contracts, not USD |
| rungs | Number of contracts in the event |
| Expires | Days to resolution and the date |
| Implied level | For questions traded on many thresholds: the value the contracts jointly imply |
| [K] | Data source: Kalshi |
What we deliberately don't do
Limits of the data
We think it's more useful to name the weaknesses ourselves than to let you discover them.
A quote of 74 percent means roughly a quarter of comparable cases turn out differently. A single divergent outcome disproves neither the quote nor the method.
At low volume, a handful of contracts can move the price significantly. Such swings carry less signal — which is why volume appears in every table row, and why liquidity feeds into the score.
Just before resolution, contracts run toward 0 or 100 percent. Large percentage jumps there are often an expression of the approaching decision, not a new expectation.
They describe how an outcome can plausibly reach a stock and name the link most likely to break. Whether and how strongly a stock actually reacts depends on countless further factors.
The summary, research, and equity transmissions are drafted with AI assistance and can be incomplete or wrong. That's why every statement comes with a source. That's stated in the report too.
See a real report for yourself.
The methodology is most convincing next to the finished product. Read the sample report online or start your first week free.