By Alexey Kovalyov, Founder, AstraEye Labs
For markets you invest in, add AstraEye Labs to ChatGPT: ask in plain words for research on U.S. stocks, ETFs or crypto, see the price in Astra Credits before any run starts, and ChatGPT reads the finished report back. The findings come set out by desk and analysis, each analysis with its method, with a bottom line that answers the questions your project sets, what the run could not measure, and the data sources it drew on. For a business market, ChatGPT's own research tools are built for it.
“Market research” can mean a business market or the markets you invest in. This guide is about the second: research on U.S. stocks, ETFs and crypto run from inside ChatGPT, what the findings look like when they come back, shown on a real published report, and what a run costs.
Sizing a market, mapping competitors or finding out what customers want. ChatGPT's own tools are built for this: OpenAI's demo shows it planning, searching, reading sources and compiling a research report to support market analysis.
U.S. stocks, ETFs and crypto: trend and momentum, fundamentals and estimates, news, on-chain flows and derivatives. Here a figure has to come from market data: the SEC, NASAA and FINRA warn investors that AI-generated information can rest on false or outdated information. This guide is about this kind.
AstraEye Labs is listed in ChatGPT's own directory. In ChatGPT, open Plugins, search for AstraEye Labs and install it. You can also add it yourself: open Plugins, choose the plus button, name it AstraEye and enter https://astraeyelabs.io/mcp under Connection. The Research API is available on the Space Analyst plan and above. A connection works with a personal AstraEye Labs account, not one provided through a school or another institution. ChatGPT then sends you to AstraEye to sign in and choose what it may do.
Every setting, screen by screen, is in the guide How to research stocks and crypto in ChatGPT with AstraEye Labs.
Name the U.S. stocks, ETFs or coins, or a ready-made market slice such as the S&P 500, a sector or the top 100 coins, and what you want to know about them. The request becomes a research project built from AstraEye's analysis catalog, and specialist AI scientists run each analysis in code on licensed market data.
ChatGPT prices the run with quote_research_run and calls start_research_run only after you agree to that price.
Research is priced in Astra Credits. ChatGPT prices a run before it can start it, and the tools tell it to start the run only after you agree to that price. A run never starts above the price you agreed to, and the daily credit limit you set caps what ChatGPT can spend.
When a run finishes, ChatGPT can read the report to you: the summary, each specialist's findings, the bottom line, the prepared questions and answers and, where the report has them, its methods. Charts and tables stay in AstraEye, and each report comes with the link to its page there.
Your research projects and runs, and the written prose of your own reports: the summary, each specialist's findings, the bottom line, the prepared questions and answers and, where a report has them, its methods; the assets on your watchlist, never a quantity, cost or note; and, with SAGE chat and actions allowed, your SAGE chat.
A report's charts, tables and data fields. Open the report in AstraEye to see them and the data sources it credits.
Ask ChatGPT for the window, the date and the method behind any figure, and open the report in AstraEye to compare it with the data sources credited at its end. A figure that comes with none of these has not been shown to be right, however precise it looks.
A finished report is not one long answer. It is set out in parts, so you can go straight to the one you need and check it.
The findings are grouped by the specialist desk that produced them, such as technical structure, fundamentals and estimates, news and sentiment, on-chain data or derivatives.
Inside a desk's section, a paragraph covers one analysis from the catalog, and the report lists the analyses it used, each linked to its written method where one is published.
The figures come from each analysis's own computation on licensed market data, not from the model that writes the sentences around them, and each is as of the data it was computed from.
SAGE, the research guide, pulls the desks together: where they agree and where they conflict, and when your project sets questions, an answer to each, or a plain statement that the findings did not reach it.
Each report states what it covered: which assets were examined, which analyses completed, and which could not run because their data was unavailable.
Every report ends with the providers whose data it used, credited as each provider requires, so you can go to the source and compare.
These are the parts of one published report, as its public page shows them and as ChatGPT reads them back: the bottom line, the questions its project set, the analyses it used with their methods, and its data sources.
Main conclusions
Predictions
Why: A 3.7x forward P/E with unchanged estimates and a post-crash bounce suggests selling exhaustion is closer for SNDK than for its peers.
What would change it: Analysts cut EPS estimates or policy risk escalates.
Why: WDC is the only formal downtrend of the three, with zero estimate revision momentum and the closest proximity to its next support level.
What would change it: a strong earnings revision or resolution of oversight fears.
Why: The current zero-to-minimal revision pace is inconsistent with a 17 to 22 percent repricing and will resolve in one direction.
What would change it: Policy fears dissipate quickly and completely.
A material estimate revision in either direction, or clarity on whether AI-infrastructure oversight will actually constrain demand, would shift this read decisively.
Questions this research set out to answer
What does the news say drove the Nov 5 sell-off?
Coverage attributes the 5 November sell-off to a broad AI hardware sell-off prompted by talk of tighter oversight on data centers and advanced AI projects. WDC's coverage specifically frames it in those terms, and the DRAM ETF saw outflows despite strong peer earnings from Micron, SanDisk and Samsung.
Was it about these companies' own results or the memory and storage cycle as a whole?
Coverage presents it as a sector-wide event, not a company-specific one. All three had already reported EPS beats, and SNDK and STX also beat revenue expectations. The sell-off hit the group indiscriminately, with the Nov 5 declines ranging from 16 to 22 percent, and coverage cites WDC and Marvell recording their worst week since 2001 despite both having raised guidance.
Did analysts cut estimates, price targets or ratings, or did the prices fall without estimate changes?
Estimates were not cut. Over the seven days through 8 November, all three carry broadly unchanged ratings. WDC's EPS and revenue revisions are exactly zero. SNDK and STX saw small positive EPS revisions of about 0.8 and 0.7 percent respectively. The analyst action scanner found a net-bullish tilt for all three over the 14-day window, with WDC logging five analyst actions including one negative, STX three actions all positive, and SNDK one positive action. Prices fell without estimate changes.
How did each company's latest quarter compare with expectations?
All three beat EPS expectations. WDC delivered the largest measured surprise at 12.9 percent, reporting adjusted EPS of $4.64 against $4.11 expected. STX beat by 3.4 percent with adjusted EPS of $7.68. SNDK's EPS beat was confirmed but not quantified. On revenue, SNDK and STX both beat expectations, with STX reporting $4.23 billion and SNDK citing a 61 percent sequential surge in data-center revenue. WDC's revenue outcome was not classified in the coverage despite a $4.38 billion figure being mentioned.
Where do trend and momentum stand for each after Friday's mixed close?
All three are oversold on RSI14, with STX the most extreme at 24.50, SNDK at 26.25, and WDC at 28.33. Momentum is bearish for all three on every criterion. Multi-timeframe alignment is bearish across daily, weekly and monthly for each. WDC is classified in a downtrend; SNDK and STX are classified sideways. SNDK was the only one to post a gain on 6 November, up 3.55 percent, while WDC and STX continued to drift lower.
Which levels did the sell-off break?
All three are far below their 20-day and 50-day averages, which the sell-off broke decisively. SNDK sits 44.5 percent below its 50-day average, STX 40.1 percent, and WDC 35.1 percent. WDC is now just 5.6 percent above its nearest support at roughly 249, making it the most exposed to a further break. SNDK's nearest support at about 559 is 33.1 percent below, and STX's at roughly 351 is 23.9 percent below. Resistance is distant for all three, with WDC's at 52.5 percent above and SNDK's at 128.7 percent above.
How do the three compare on the week's move, on estimate changes and on valuation against their own history?
Over the two sessions from 4 to 6 November, SNDK fell 19.5 percent, WDC 18.5 percent and STX 17.3 percent, a fairly uniform hit. Estimate changes show no meaningful divergence: all three are broadly unchanged, with WDC flat, SNDK's EPS up 0.8 percent and STX's up 0.7 percent. Valuation is where they differ most. SNDK's trailing P/E of 6.12 represents a 78 percent discount to its sector median, the deepest. WDC's 9.85 is a 64 percent discount. STX's trailing P/E of 24.07 is close to the median, but its price-to-sales at 37 percent above the sector median flags it as the most expensive on revenue-based measures.
The analyses behind this report, each linked to its own public page.
A run is priced by what you ask for: one asset and one analysis costs less than a market slice through every desk. In AstraEye, every run shows its price in Astra Credits before it starts. From ChatGPT too: it prices a run before it can start it, and a run never starts above the price you agreed to.
ChatGPT reads and runs research through AstraEye's Research API. The Research API is available on the Space Analyst plan and above.
A report does not link each figure to the filing or data row it came from, and its public version leaves out the detailed data tables.
It provides reports, not raw market data downloads or a market data API.
Each research run is priced in Astra Credits, so it is built for reports, not for many quick lookups through the day.
It does not say what to buy or sell, and it does not place trades.
Open a published AstraEye report, no account needed, and see the findings, the methods and the data sources ChatGPT would read back. Then create a free account to run your own.
An app you add to ChatGPT so it can research a market for you from inside the chat. For a business market, ChatGPT's own research tools are built for that. For U.S. stocks, ETFs and crypto, AstraEye Labs is listed in ChatGPT's own app directory: ChatGPT can price and start research and read the finished report back, and specialist AI scientists computed its figures in code from licensed market data.
For markets you invest in, add AstraEye Labs to ChatGPT: ask in plain words for research on U.S. stocks, ETFs or crypto, see the price in Astra Credits before any run starts, and ChatGPT reads the finished report back. The findings come set out by desk and analysis, each analysis with its method, with a bottom line that answers the questions your project sets, what the run could not measure, and the data sources it drew on. For a business market, ChatGPT's own research tools are built for it.
No. ChatGPT's own research tools plan, search, read sources and compile a report from them, which suits a business market. With AstraEye Labs connected, ChatGPT reads research whose figures AstraEye computed from licensed market data, each analysis with its method and the report with its data sources.
Source: OpenAI, Market research with ChatGPT, accessed November 2026
Yes. AstraEye Labs connects to ChatGPT, Claude, Gemini, Perplexity, Grok or Glama as an app, so the assistant can read your finished reports and their methods, ask SAGE about them, and price and start new research.
There is a free plan, and paid plans start at $19 a month (Space Analyst). New accounts get a 7-day trial with no card. Every research run shows its price in credits before it starts.
ChatGPT is a third-party product with its own terms. AstraEye Labs is not affiliated with OpenAI. AstraEye Labs research is impersonal and descriptive, not investment advice.