Kalshi Earnings Call Markets: AI Research Workflow

Kalshi's company-report markets are turning earnings season into something more specific than "will the stock go up?"

The new question is:

What will management actually say on the earnings call?

That is a different research problem. It is not just EPS, revenue, or guidance. It is language. It is whether a CEO mentions a product, a region, a supply-chain issue, a competitor, a partnership, a tariff, a macro risk, or a strategic priority while answering analyst questions.

That is exactly where earnings call transcripts and AI analysis matter.

Stock Alarm Pro already gives investors a way to read earnings call transcripts, generate AI summaries, and ask Atlas to analyze what management said. For company-report prediction markets, that same workflow becomes a research edge: study the prior calls, identify the recurring language, then compare the market's implied odds to what the transcript history actually suggests.

This is not a guarantee. Prediction markets are risky, contracts can expire worthless, and liquidity matters. But it is a much better process than trading a topic market from memory.

The Kalshi Company-Report Example

Kalshi's McDonald's company-report page shows the new format clearly.

For McDonald's Q2 2026 report, Kalshi lists a company page with earnings timing, consensus EPS, and multiple market types. The most interesting section asks:

What will McDonald's say during their next earnings call?

The page shows topic contracts tied to words or themes such as:

• World Cup
• Supply chain
• Red Bull
• Iran / Middle East
• McCafe
• Loyalty
• Competition
• Happy Meal
• Big Arch
• Delivery
• Snack Wrap

That is a fundamentally different game from simply asking whether MCD beats EPS.

For a topic contract, the question is usually not "is this important?" The question is:

1. Has management used this language before?
2. Is the topic relevant to the quarter?
3. Does the company have an incentive to bring it up?
4. Do analysts usually ask about it?
5. Is the market price too high or too low versus that historical base rate?

Those are transcript questions.

Why Prior Earnings Calls Matter

Most earnings calls follow patterns. Management opens with prepared remarks. The CFO covers the numbers. Analysts ask about the same recurring drivers: demand, margins, pricing, promotions, competition, capex, labor, inventory, international markets, and guidance.

If you are researching whether a company will mention a topic, prior calls give you three useful baselines.

1. The Recurrence Baseline

Some words appear almost every quarter.

For a restaurant company, words like "loyalty," "delivery," "value," "traffic," "franchisee," and "margin" may be part of the normal operating vocabulary. A market priced at 70% on a word that appeared in 9 of the last 10 calls may be very different from a market priced at 70% on a word that appeared once.

The first step is not cleverness. It is counting.

Open the prior calls and ask:

• Did the topic appear in prepared remarks?
• Did it appear only in Q&A?
• Did it appear because an analyst asked about it?
• Did it appear across several quarters, or only during one event?

2. The Catalyst Baseline

Some words only appear when a specific catalyst exists.

For example, a company may mention a product launch only near rollout. It may mention tariffs only when the policy environment changes. It may mention the Middle East only when there is a supply-chain, franchise, commodity, or geopolitical reason to do so.

This is where transcript history and current news need to be combined. If a topic never appears in normal quarters but suddenly has a fresh catalyst, the prior base rate may understate the probability.

3. The Analyst-Question Baseline

Some topics are not volunteered by management. They appear because analysts ask.

That matters because a word market can resolve based on the full call, not just prepared remarks, depending on the contract rules. If analysts reliably ask about same-store sales, loyalty, promotions, or a new product, that can raise the probability even if management would not lead with it.

Before trading a "will they say X" market, you want to know whether X is management-driven or analyst-driven.

How Stock Alarm Pro Helps

Stock Alarm Pro has a dedicated earnings page for each covered stock.

For McDonald's, the workflow starts at MCD earnings. For any covered ticker, open:

/quote/TICKER/earnings

That page gives you:

• Recent earnings call transcripts
• An AI earnings summary card
• Key themes from management commentary
• Takeaways from the latest call
• Guidance and risk language where available
• A structured starting point before you read the full transcript

Then you can use Atlas AI to ask follow-up questions.

Instead of asking a vague question like "will McDonald's mention delivery?", ask questions that force a transcript-based answer:

text code-highlight
Look at MCD's recent earnings calls. How often does management mention delivery,
loyalty, McCafe, Snack Wrap, Big Arch, and World Cup? Separate prepared remarks
from analyst Q&A where possible.

Or:

text code-highlight
For MCD, which topics have appeared consistently in recent earnings calls, and
which only appeared around a specific product, geography, or promotion catalyst?

Or:

text code-highlight
Compare the latest MCD earnings call themes to the prior four quarters. Which
topics are becoming more prominent, which are fading, and which are one-off?

That is the right posture. AI is not "picking a bet." AI is compressing a transcript stack into a research map.

The Research Workflow for Topic Markets

Here is the simple process.

Step 1: Read the Contract First

Before opening a transcript, read the market rules.

You need to know:

• What exact word or phrase counts?
• Does it include prepared remarks and Q&A?
• Are plurals, abbreviations, brand spellings, or translations included?
• What source determines resolution?
• What happens if the call is cancelled, delayed, or transcribed differently?

This is not a small detail. A topic can be directionally right and still lose if the exact resolution language is narrower than you assumed.

Step 2: Build the Transcript Base Rate

Open the prior earnings calls in Stock Alarm Pro and check how often the topic appeared.

For each topic, create a quick table:

The exact counts matter less than the classification:

• Always-on language - common every quarter
• Recurring operating topic - appears often but not always
• Catalyst topic - appears when a specific event exists
• Analyst-driven topic - appears mostly in Q&A
• Low-base-rate topic - rare unless there is a fresh reason

Step 3: Compare Base Rate to Market Price

A prediction market price is an implied probability.

If a contract trades at 69 cents, the market is roughly saying 69% before fees and spread. Your job is not to decide whether the topic is interesting. Your job is to decide whether the probability is wrong.

For example:

• If a topic appeared in 8 of the last 8 calls and is central to the company's strategy, a 55% market may deserve attention.
• If a topic appeared once in the last 8 calls and has no current catalyst, an 80% market may be too rich.
• If a topic has low historical recurrence but a major current catalyst, history alone may be too bearish.

This is where AI helps most. It gives you the transcript map faster, but you still need to think.

Step 4: Check Current Quarter Context

Prior calls are the base rate. The current quarter is the adjustment.

For company-report markets, check:

• The latest earnings date and report timing
• Consensus EPS and revenue expectations
• Segment trends
• Recent product launches
• Macro or geopolitical news
• Analyst preview notes where available
• Recent company press releases
• Whether the topic is likely to be asked by analysts

Stock Alarm Pro helps here because the earnings page sits next to the quote page, AI analysis, fundamentals, news, analysts, and chart context.

You do not need ten tabs just to answer "what has this company been talking about?"

Reported-Metric Markets Are Different

Kalshi company-report pages may also include markets on reported metrics, not just call language.

The McDonald's page, for example, includes markets tied to quarterly and full-year figures such as comparable sales growth or systemwide restaurants.

Those markets require a different workflow:

• Historical financials
• Segment reporting
• Consensus estimates
• Company guidance
• Street preview context
• Seasonality
• Prior-quarter trend

Earnings call transcripts still matter, but they are not enough. For reported metrics, management commentary is one input. The actual financial model matters more.

That is where Stock Alarm Pro's quote pages, fundamentals, analyst pages, and Atlas summaries can be paired with the transcript workflow.

What AI Should and Should Not Do

AI is useful for:

• Summarizing long earnings calls
• Extracting recurring topics
• Comparing quarter-over-quarter language
• Separating prepared remarks from Q&A
• Finding risk, guidance, and product commentary
• Creating a research checklist before the call

AI should not:

• Ignore the exact contract rules
• Pretend a transcript pattern guarantees a resolution
• Trade based on a summary alone
• Skip liquidity, fees, bid-ask spread, and position sizing
• Treat a 70% probability as "certain"

The right workflow is human-led, AI-assisted.

Let AI do the tedious transcript compression. Let the market price tell you what other traders believe. Then decide whether the evidence is strong enough to justify taking risk.

A Practical MCD Research Prompt

If you are looking at a McDonald's company-report market, this is the kind of Atlas prompt to use:

text code-highlight
Analyze MCD's recent earnings call transcripts for topic-market research.
For World Cup, supply chain, McCafe, loyalty, competition, Big Arch, delivery,
and Snack Wrap, classify each as always-on, recurring, catalyst-driven,
analyst-driven, or low-base-rate. Include the reasoning and what current-quarter
context would increase or decrease the probability.

Then follow with:

text code-highlight
Which of these topics appeared in prepared remarks versus Q&A in recent calls?
Which ones are likely to be mentioned only if analysts ask?

That second question is often where the edge is. A topic that management never volunteers but analysts ask every quarter may still be more likely than it looks from prepared remarks alone.

The Takeaway

Company-report markets make earnings research more granular.

It is no longer only:

• Did the company beat EPS?
• Did revenue beat?
• Did the stock gap up?

Now traders can express views on what management will say, what metrics will print, and which themes will show up on the call.

That makes earnings call transcripts more valuable, not less. If the market is pricing whether a word or theme appears, the prior transcript record is the first thing to study.

Stock Alarm Pro gives you that starting point: transcripts, AI earnings summaries, Atlas follow-up analysis, quote pages, fundamentals, news, and alerts around the same ticker.

Use it to build a better research baseline. Then size risk like the market can still surprise you.

Related Reading

• How to Read an Earnings Calendar
• How to Set Up Earnings Alerts So You Never Miss a Report
• Prediction Market Analysis
• AI Investing Guide
• MCD Earnings Page
• Atlas AI Research