Skip to content

AI stock analysis you can inspect.

Ask a financial question, examine the evidence, and keep the work behind the answer. Ghost brings company research, market data, and backtesting into the same assistant as your notes and next steps.

Source-led research · Testable assumptions · Private beta for Mac

Stock researchIllustrative workflow

Atlas — company research brief

Question
What changed since the last earnings report?
Evidence
Company filing, earnings transcript, dated price history
Analysis
Business drivers, valuation assumptions, downside questions
Next step
Save the brief. Verify the open questions.
Source contextWork you can review

Keep the question, data, and assumptions together.

An illustrative company review shows how financial research can become a source-linked note, a reproducible calculation, and a later review task.

Ask a question that can be investigated.

Specify the company, period, and what you want to understand. Distinguish an evidence question from a request to predict the next price move.

Company review · Atlas

What changed after the latest earnings report?

Evidence
Company filing and earnings transcript
Comparison
Earlier reported period
Output
Sourced brief with unresolved questions

Keep reported facts separate from interpretation.

Read the relevant source material and record its date. Current market data and older saved context must not be silently treated as the same thing.

Evidence check
Reported
What the source says
Inferred
What the evidence might imply
Missing
What still requires verification

Make the calculation inspectable.

Use the trading workspace for supported data analysis or a historical test. Retain the dataset, period, rules, and costs so the conclusion can be reviewed.

Analysis record
Data
Source, window, interval, and missing values
Method
Calculation or simulation rules
Limits
Costs, bias, and sensitivity
Simulation, not a trade

The workspace does not itself authorize live orders.

Save the thesis and the next question.

Keep the conclusion and its sources in a note. Create a task for the evidence gap, or configure a future research run with its own brief and timezone.

Research handoff
Note
Thesis and references
To-do
Verify the remaining assumption
Scheduled task
Revisit the specified question later
Explore scheduled research. Results and status are reviewed in Scheduler.

Illustrative research only, not investment advice or a performance result. Data may be delayed or incomplete. Historical simulation does not predict future returns.

Research, with the working behind it.

AI stock analysis uses software to help examine company information and market data. Ghost can gather sources, compare evidence, run calculations in a trading workspace, and explain the result. The useful output is a research record you can inspect—not a promise about where a share price goes next.

Company research starts with the source.

Ask Ghost to investigate a business, compare companies, or work through a specific earnings question. A useful brief separates what the company reported from an assumption, a forecast, or the assistant’s interpretation.

Bring the filing, transcript, date, and relevant passage into the analysis. When sources disagree or information is missing, keep that uncertainty visible instead of forcing a confident conclusion.

Use multi-source research to follow a question beyond one page. Save the evidence and your own thesis in a working research note so the next review has a starting point.

Reported
What the company actually disclosed.
Inferred
What the evidence might imply.
Unresolved
What still needs a source or a better explanation.

Put market data to work.

The trading workspace gives Ghost tools for working with price history, dataframes, indicators, and quantitative analysis. Ask for a chart, a comparison over a defined period, or a calculation with the assumptions written down.

Fundamental analysis and technical analysis answer different questions. A business explanation should not quietly turn into a trading signal, and an indicator should not become evidence of a guaranteed outcome.

Coverage, delays, and availability depend on the data source. Ask for the source, observation time, currency, period, and any missing data alongside the chart. A saved dataset is not a live feed.

Input
Dataset and source identified
Window
Period, interval, and currency stated
Output
Chart, calculation, and limitations together

Backtest the idea. Keep the assumptions.

Use backtesting to explore how a defined strategy behaves on historical data. Ghost’s trading sandbox includes tools for simulation and quantitative work, with workspace files you can revisit.

Make the test inspectable: specify the universe, period, entry and exit rules, costs, and data treatment. An attractive historical result is a reason to examine the method—not a guarantee that it will work in the future.

The managed sandbox is a place to run the analysis. It does not ship broker credentials or authorize live orders. Research and simulation remain separate from any trading decision.

Define
Rules and historical window
Test
Calculate and compare outcomes
Review
Costs, bias, gaps, and sensitivity

Return to the thesis, not a blank chat.

Keep the research question, supporting files, and conclusion connected. A note can hold the thesis; a to-do can hold the question that still needs work. Relevant earlier context can help Ghost pick up the investigation.

For a future review, configure a scheduled research task with a clear brief, cadence, and timezone. Review its output and status in Scheduler. Scheduling a run does not itself guarantee a notification or a continuously running market monitor.

When you need to explain the findings, turn the reviewed material into a presentation. Keep sources and assumptions with the story instead of losing them between tools.

Retain
Thesis, sources, and analysis files
Revisit
A specific question at a chosen time
Explain
A reviewed brief or editable presentation

Product details

Stock research, at a glance.

Supported work, useful outputs, and what to check before you start.

Inputs
A company or market question, a period, and the available financial sources. The chosen data provider determines coverage and freshness.
Outputs
Company research, charts, and historical tests can become a working note or presentation, with the source and assumptions kept alongside the analysis.
Interpretation
Historical performance is not a forecast. Review data gaps, assumptions, and test conditions; a research result is not a promise of trading returns.

One assistant. Connected work.

Ghost is a personal AI assistant for Mac. Stock research is one capability of the same product.

A company question can move through Research, run calculations in a Sandbox, become a Note, and leave a To-do for the next review. These are capabilities of Ghost, not separate products or automatic steps you must use together.

  • Research

    Collect evidence beyond a single source.

  • Notes

    Keep the thesis and its references.

  • Sandboxes

    Run and retain the analysis.

Questions about stock research.

Is this an AI stock picker?

Ghost is built to help with research and analysis. It can investigate a company or test an idea, but this page does not promise stock rankings, winning trades, or a reliable prediction of future prices.

Is the market data always real time?

No. Availability and freshness depend on the selected source. An analysis should identify its source and time window, and flag missing or delayed data.

Can Ghost backtest a strategy?

Ghost can use the trading workspace for supported historical simulations. The rules, dataset, costs, and assumptions need to be explicit. Past performance is not a guarantee of future results.

Does the trading sandbox place orders?

The sandbox provides a research and command-execution workspace. It has no built-in broker credentials and does not itself authorize live trades.

Can I revisit the same research later?

Yes. Keep the thesis in a note and retain the supporting workspace files. You can also configure a scheduled task for a future review, with its output and status available in Scheduler.

How do I get access?

Ghost is in private beta for macOS. Request access from the homepage. Research outputs are informational and may contain errors; they are not personalized investment advice.

Keep the evidence behind the answer.

Research the question, test the assumptions, and carry the result into the rest of your work. Ghost is in private beta for macOS.

Request access