Claude and ChatGPT Prompts for Indian Stock Analysis, Backed by Real Data

The best Claude and ChatGPT prompts for Indian stock analysis do two things: they ask for specific metrics over specific periods, and they run against real NSE and BSE data instead of the model's memory. This page collects prompts you can copy, grouped by task, with the Tapetide MCP tools each one calls. Connect your assistant once (Claude.ai, ChatGPT, Claude Code or any other MCP client) and every prompt below returns numbers with dates you can check.

#Why the same prompt gives different answers

Ask a plain chatbot "analyse Maruti Suzuki fundamentals" and you get a confident paragraph built from training data. Revenue may be two years old, the margin may predate a price cut, and nothing tells you which. That is the main failure of prompt lists on the web: they assume the model already knows the numbers.

With an MCP connector, the assistant calls a tool, reads the result and quotes it. The Tapetide MCP server exposes 48 tools over about 8,000 NSE and BSE listed companies, covering financials, shareholding, price history, technicals, filings, FII/DII flows and options. The prompts below are written for that setup. Each one:

• names the company by NSE symbol, or asks the assistant to resolve it first with search_stocks
• asks for exact metrics and periods
• asks for a table
• asks for the reporting period or observation date of every number

That last habit matters most. Quotes, results and shareholding each update on their own schedule, and a dated number is one you can verify on the stock page.

#Fundamental analysis prompts

#Eight-quarter fundamentals check

For MARUTI, show the last 8 quarters of revenue, EBITDA margin, net
profit and EPS in a table, in crore. Add year-on-year growth for revenue
and profit. State the quarter for each column and flag any quarter where
profit fell more than 20% year on year.

Tools: search_stocks, then get_financials. If the assistant fills a missing cell with an estimate, ask it to leave the cell blank and say the value was not returned.

#Balance sheet and cash flow quality

For ASIANPAINT, compare operating cash flow with net profit for the last
5 financial years. Show debt to equity, interest coverage and working
capital days for the same years. Where cash flow is below profit, say by
how much.

Tools: get_financials with annual statements. A company that reports profit but not cash is worth a closer look at receivables.

#Peer comparison

Compare HDFCBANK, ICICIBANK, KOTAKBANK and AXISBANK on P/E, P/B, ROE,
net interest margin and gross NPA. One row per bank, latest available
values, with the period or date of each figure.

Tools: get_batch_quotes for valuation (one call for up to 20 stocks) and get_financials per bank. Asking for one row per company keeps the table easy to scan.

#Estimates against results

For INFY, show analyst revenue and EPS estimates next to reported results
for the last 4 quarters. Did the company beat or miss each time, and by
what percentage?

Tools: get_forecasts. Where coverage is thin, the tool returns fewer analysts or no estimates; the assistant should say so.

#Technical analysis prompts

The data behind these is end-of-day, which suits positional and swing analysis.

#Trend and momentum snapshot

For LT, give the latest close, the 50-day and 200-day moving averages,
RSI and MACD, and the 52-week high and low. Is the price above or below
each average? Include the date of the last close.

Tools: get_company_profile with include: ["technicals"] and get_stock_quote.

#Volume and delivery

Pull 6 months of daily prices for TRENT with delivery percentage. List
the 10 sessions with the highest delivery percentage and the price change
on each of those days.

Tools: get_price_history, which returns daily open, high, low, close, volume and delivery percentage.

#Technical screen

Find Nifty 500 stocks trading above their 200-day moving average with
RSI between 40 and 55 and volume above the 20-day average. Show symbol,
close, RSI and sector, sorted by RSI.

Tools: screen_stocks_technical. For more screening patterns, see the stock screener MCP guide.

#Ownership and governance prompts

#Who is buying

Show the shareholding pattern of POLYCAB for the last 6 quarters:
promoter, FII, DII and public. Highlight any change above 1 percentage
point and list the mutual funds holding it.

Tools: get_shareholding and get_stock_ownership.

#Pledge, deals and ratings

For ADANIPORTS, list promoter pledge by quarter, any bulk, block or
insider deals in the last 90 days, and credit rating actions in the last
2 years. Dates for everything.

Tools: get_promoter_pledge, get_stock_deals and get_credit_ratings. This is a fast governance check before you spend time on a mid cap.

#Score with its weak spots

What is the Tapetide Score for DIXON? Show the six pillar scores, the
data confidence and any governance flags. Which pillar pulls it down
most?

Tools: get_tapetide_score. The method is on the score methodology page, and monthly score changes are in the Tapetide Score report.

#Earnings calls and filings prompts

#Management commentary

Summarise the latest earnings call for BAJFINANCE. What guidance did
management give, what risks did they mention, and what did analysts push
on? Note the call date.

Tools: get_earnings_call_summary.

#Read the annual report itself

List the parsed filings for TITAN. Open the latest annual report and
quote what it says about store expansion plans, with page numbers.

Tools: list_company_documents, then read_document (up to 12 pages per call, with page markers to cite). Asking for page numbers lets you open the PDF and check the quote.

#Market context prompts

How have FII and DII net flows moved over the last 20 trading days? Show
weekly totals and India VIX at the end of each week.

Is the Nifty 50 cheap or expensive versus its own history on P/E and P/B?
Give the current values, the 10-year range and the observation date.

Which sectoral indices gained the most over the last three completed
weeks? Show returns for the top 5 and bottom 5.

Tools: get_fii_dii_detail, get_india_vix, market_valuations and get_index_performance. The FII/DII guide has more flow prompts, and the monthly FII/DII report shows the same data as a page.

#Turn a prompt into a routine

Prompts are most useful when you reuse them. A quarterly results check, for example, is the same four questions for every company you follow: did revenue and profit grow, did margins hold, did the cash flow match the profit, and did promoters or institutions change their holdings. Save those as one prompt with a placeholder for the symbol and run it after each results season.

In Claude, a Project with these prompts in its instructions keeps them one click away. In Claude Code or another coding agent, you can ask for the output as a markdown note per company, so next quarter you can ask "compare this quarter with last quarter's note" and see what moved. In ChatGPT, keep them in a pinned chat. Whichever client you use, start each routine by asking the assistant to call read_me, which returns the current tool guide and does not use your allowance.

#Three habits that fix most bad answers

1. Ask which tools ran. "Which tools did you call for this answer?" shows you whether a number came from the connector or from the model's memory. Treat anything not from a tool as unverified.
2. Resolve the company first. Many Indian groups have similarly named listed companies. "Confirm the NSE symbol and ISIN before you start" avoids analysing the wrong one.
3. Keep each prompt to one job. Ten companies across eight quarters plus ownership plus filings in one prompt burns calls and invites mixed-up periods. Split it.

#Limits

• Not a live trading feed. Quotes are the latest available prices, results and shareholding follow reporting periods, and index option chains are end-of-day snapshots.
• Free plan: 50 successful tool calls a day, no monthly cap. Failed calls do not count. Higher limits are on the pricing page.
• TapeAi is separate. Tapetide's own assistant at /chat has its own allowance. The MCP allowance covers Claude, ChatGPT and other clients you connect.
• Not advice. The assistant can still misread a table or overstate a trend. Check important figures on the stock page.

For a side-by-side view of AI research tools for Indian stocks, see best AI stock analysis apps in India. Developers who want the same data in code can start at the Tapetide developer docs.