I Built a Claude Skill That Writes Equity Research Reports

1 Click to report

A junior analyst takes 5 days. This takes 12 minutes. The output is 90,000 characters of professional-grade research.

Let me tell you what happened when I ran it on TD Power Systems last week.

I typed one command. Twelve minutes later, I had an 8-tab, interactive HTML

Complete with a 16-point fraud screen, Porter’s Five Forces, a 3×3 scenario valuation matrix, a DCF sanity check, peer comparisons against WEG Brazil and Andritz Austria, and a Tankrich Quality Score of 65/100 with a detailed breakdown of exactly why.

The report told me the stock at ₹882 is pricing in FY28 earnings at 35–50x. It flagged that the promoter stake had declined 31.6% over three years without clear explanation. It computed that the OCF/PAT ratio had collapsed to 0.23x in FY25. It calculated the incremental ROIC at 29.7% and graded capital allocation an A.

I didn’t write any of that manually. I built a skill that did

The Problem I Was Solving

I’ve spent years building the Tankrich Investment Framework — multiple analytical sheets covering everything from business model quality to fraud detection to scenario valuation. It’s the framework I use for every stock I research seriously.

The problem? Running it properly on one company takes a trained analyst many days. Ratios to compute. Qualitative questions to answer. Competitor pages to fetch. Management commentary to parse. Scenario matrices to build. Charts to draw.

Most retail investors don’t have 3–5 days. Most small advisory firms don’t have a junior analyst to spare. And most people who want to do serious research get halfway through and give up because the process is too exhausting.

So I asked myself: what if the framework itself could run automatically?

What I Built

The Tankrich Stock Analyst is a Claude skill a structured set of instructions that tells Claude exactly how to conduct a full equity research analysis on any NSE/BSE-listed company.

It runs in three sequential passes:

Pass 1 — Data Collection. It fetches financials from multiple sources, runs 6–8 targeted web searches, discovers 2–3 Indian and 2–3 global peers, and computes 19 quantitative metrics from ROIC to Beneish M-Score signals to the Buffett $1 Test.

Pass 2 — Qualitative Analysis. It applies all Tankrich Framework sheets. Every question answered. Every ratio interpreted in context. Porter’s Five Forces with real competitor names. Fraud screen with 16 individual items. Capital allocation scorecard. Moat detection. Value migration classification.

Pass 3 — Report Assembly. It builds a single self-contained HTML file with 8 tabbed sections, 7+ interactive Chart.js charts in a green-and-white editorial theme, and a 3×3 scenario valuation matrix with target prices at two PE ranges. The whole thing opens in any browser and prints to PDF.

The output is 80,000–100,000 characters.

Why This Changes Something

Here’s what I keep coming back to: the bottleneck in investment research was never intelligence. It was time.

Knowing how to analyse a company — understanding what OCF/PAT means, knowing to check if receivables are growing faster than sales, knowing to look at incremental ROIC and not just headline ROE — that knowledge is teachable. The Tankrich framework encodes it.

The bottleneck was the grinding execution work. Pulling numbers. Cross-referencing years. Fetching competitor data. Building the table. Writing the narrative. That work ate hours. It’s why most retail investors don’t do it, why small advisors rely on recycled brokerage research, and why deep fundamental analysis has always been a resource available mainly to institutions.

Now it’s a 12-minute task.

What It Catches That Brokerage Reports Don’t

Brokerage research is optimized for different things than yours and my interests suggest. Target prices are often anchored to recent momentum. Risks are listed but not seriously weighted. Governance issues are politely mentioned then forgotten.

The Tankrich Analyst doesn’t have those incentives.

On TD Power Systems, it flagged the promoter stake decline of 31.6% over three years — a data point that was publicly available but that I hadn’t seen prominently highlighted in any brokerage note on the stock. It computed that the FY25 OCF/PAT of 0.23x was the weakest in five years and built an explicit monitoring threshold: if OCF/PAT stays below 0.5x for three consecutive years, the earnings quality thesis breaks.

It also told me the company’s asset-based floor value is just ₹21 per share — 2.4% of the ₹882 CMP. Meaning the entire investment is a bet on sustained future earnings, with zero balance sheet protection. That’s not a reason not to invest. But it’s a fact every investor should know going in.

Brokerage research rarely says that as plainly.

Who Should Use This

• If you’re a SEBI-registered RIA managing client portfolios without a full research team, this gives you institutional-depth coverage on any stock, on demand, in minutes — with your own analytical framework stamped on it.

• If you run a PMS or family office, this becomes your first-pass research engine. Your analysts stop spending days pulling together raw data and start spending their time on the 20% that actually requires human judgment — management calls, reading between the lines, forming a conviction view.

• If you’re an independent analyst or financial content creator, this is the difference between covering 5 stocks a month and covering 25.

• If you’re a serious retail investor who has always wanted to do real fundamental research but found it too time-consuming — this is what you’ve been waiting for.

What It Doesn’t Do

I want to be honest about the limits, because hype without honesty is how trust gets destroyed.

The skill works from publicly available data. It cannot replace a conversation with management. It cannot attend an investor day. It cannot pick up on the tone of a CFO’s voice when they answer a question about receivables. It cannot read unpublished supply-chain data or channel-check distributors.

It produces the 80% of analysis that’s based on structured data and systematic frameworks. The remaining 20% — primary research, qualitative judgment, edge from information asymmetry — that’s still on you.

Think of it as the most thorough, most consistent research associate you’ve ever had. One who never gets tired, never skips a step, and applies the same rigour to the 47th company as to the 1st.

Also it is unlikely to work on banks and financial services business

A Note on the Framework Itself

The Tankrich Investment Framework didn’t happen overnight. It’s the accumulated product of years of reading Buffett letters, Charlie Munger’s mental models, Michael Porter’s competitive strategy work, the writings of practitioners who’ve actually managed money across market cycles.

The sheets encode specific things I’ve learned matter: that incremental ROIC is a better management quality signal than headline ROE, that fraud detection requires looking at 16 specific items not just “governance,” that valuation without a scenario matrix is just guessing, that the Earning Power Box tells you something about a business’s quality that no single ratio can.

Encoding that framework into a skill means it gets applied consistently. No shortcuts. No “I’ll skip the cash flow analysis because the margins look great.” Every company gets the full treatment, every time.

The Bigger Picture

We’re at an inflection point in financial analysis. The tools that were previously only available to analysts at top-tier funds systematic frameworks, global data access, rapid synthesis are now available to anyone willing to build with them

The Tankrich Stock Analyst is my attempt to make sure the leveling up is accessible.

"If you want a sample report on a stock of your choice, reply with the ticker. First 5 are free."