11.1 · Idea generation¶

Why this matters: analysis is expensive; ideas are cheap, but good ideas are not. A deep dive takes weeks, so what you choose to dive into decides your returns more than how well you dive. This lesson is about building a funnel — many candidates in, few survivors out — from screens, events, insider behaviour, sector work and reading, inside a circle of competence you can actually defend.

Learning objectives — after this lesson you can:

Run quantitative screens (quality, value, growth, magic formula, coffee-can, net-net) on Screener.in with concrete queries, and know what each finds and misses.
Use event-driven sources: 52-week lows/highs, corporate actions (spin-offs, demergers, buybacks, open offers), insider and promoter buying, index changes.
Generate ideas from sector work, reading and other investors' disclosures.
Define your circle of competence and keep a funnel and watchlist with kill criteria.

Prerequisites: 04.6 Ratio dashboard, 06.5 Multiples · Time: ~70 min

1. The funnel¶

flowchart LR
  S[Sources<br/>screens, events, insiders,<br/>sectors, reading] --> Q[Queue<br/>one line each]
  Q --> K[Kill test<br/>30 min]
  K --> T[Triage<br/>3 hours]
  T --> D[Deep dive<br/>2–3 weeks]
  D --> M[Memo & decision]
  K -.->|~80% die| X[Kill log]
  T -.->|~60% die| X

Base rates from practitioners: of 100 ideas queued, 15–20 survive a kill test, 5–8 survive triage, 2–3 become positions. A funnel that does not kill is not a funnel. The kill log — why each idea died — is how you learn whether your sources are any good.

2. Quantitative screens¶

Screens are hypotheses about where mispricing lives, expressed as filters. Screener.in's query language (free with an account) is the practical tool in India; the queries below run as written (adjust thresholds).

Screen	Hypothesis	Screener query (illustrative)	Finds	Misses / traps
Quality	High, stable returns on capital persist and are under-priced by impatient investors	`Return on capital employed > 20 AND Average return on capital employed 5Years > 18 AND Debt to equity < 0.5 AND Sales growth 5Years > 10 AND Market Capitalization > 500`	Compounders; moated businesses	Usually expensive; peak-cycle ROCE for cyclicals; ignores governance
Value	Low multiples mean-revert	`Price to Earning < 15 AND Price to book value < 2 AND Return on equity > 12 AND Dividend yield > 1`	Cheap, profitable, paying companies	Value traps (cheap for a reason); PSUs and cyclicals dominate; peak earnings
Growth at a reasonable price	Growth not fully priced	`Sales growth 3Years > 15 AND Profit growth 3Years > 15 AND Price to Earning < 25 AND Return on capital employed > 15`	Mid-caps in early re-rating	Growth bought with capital or credit; check CFO
Magic formula (Greenblatt)	Rank by earnings yield and ROIC together	Compute EBIT/EV and EBIT/(NWC + net fixed assets); rank each; sum ranks; top 30	Cheap and good	Needs your own EV/ROIC computation; excludes financials/utilities by design
Coffee can (Saurabh Mukherjea's Indian version)	Consistent growers held for a decade	`Sales growth 10Years > 10 AND Average return on capital employed 10Years > 15` with a year-by-year consistency check on the export	Consistent compounders	Survivorship; fully priced; the 2021–24 experience shows multiples matter
Net-net (Graham)	Below liquidation value	`Market Capitalization < (Current assets − Total liabilities)` — build from the export; Screener's fields differ	Deep distress; tiny companies	In India, IBC waterfall leaves equity nothing; often governance disasters; illiquid
Cash-flow quality	CFO backs profit	`Cash from operations last year / Net profit last year > 0.8 AND Return on capital employed > 15 AND Debt to equity < 1`	Honest profits	—
Dividend + buyback yield	Shareholder returns as a signal of surplus cash	`Dividend yield > 2 AND Return on equity > 15` plus buyback history	Cash-rich, disciplined	Ex-growth; PSUs
Piotroski high-F, low-P/B	Improving fundamentals in cheap stocks	Build with `fi.forensics.piotroski_f` on a Screener export; P/B < 1.5, F ≥ 7	Turning value names	Governance overlay essential (09.5)
Forensic exclusion	Remove the dangerous	Exclude: promoter pledge > 20%; CFO/PAT 3-yr < 50%; auditor resignation; ASM/GSM	—	Run on every list, not as an idea source

Practical rules: screen on three-to-five-year averages not single years; export to Excel and add columns the screener lacks (CFO/PAT, pledge, RPT share); rerun monthly and watch entrants and exits rather than the static list; and remember that every screen is run by thousands of others — the edge is in what you do next, not in the query.

3. Event-driven sources¶

Source	Why it produces mispricing	Where
52-week lows (with reason)	Forced or emotional selling; the question is whether the reason is permanent	Screener/exchange lists; check the news and the numbers before the narrative
52-week highs / breakouts	Momentum with fundamentals (post-earnings drift)	Same; fundamental confirmation needed
Demergers and spin-offs	Spun entities are sold by index funds and holders who didn't want them; get their own multiples later	Scheme announcements (exchanges); NCLT approvals; 12.3
Buybacks (tender offers)	Signal of surplus cash and management's price view; arbitrage in the acceptance ratio	Announcements
Open offers / delisting	Price floors and premiums	SAST filings
Rights issues	Renunciation mispricing; forced participation	Announcements
Promoter / insider buying	Informed buyers	PIT Reg 7 disclosures; SAST creeping acquisition
Institutional 13F-like data	India: mutual-fund monthly portfolios (AMFI/AMC sites), bulk/block deals, superstar-investor holdings (shareholding pattern > 1%)	Follow changes, with a lag discount
Index inclusion/exclusion	Forced flows	Index-provider announcements (12.3)
IPOs / QIPs / OFS	Supply events; pricing set by bankers	SEBI DRHP list; underperformance base rates (03.5)
Regulatory changes	Winners and losers repriced slowly	SEBI/RBI/ministry circulars; budget documents
Results season	Post-earnings-announcement drift; guidance changes	Exchange filings; concalls (03.4)

4. Sector rotation and reading¶

Sector work (Module 08): a sector map yields the two or three companies best placed for the next phase (the capital-cycle turn; the regulatory change); ideas from structure, not screens.
Competitors' concalls: the best free source on any company is its rival's call; also the source of "who is taking share".
Reading widely: annual reports of companies you don't own (Buffett's method); trade journals; the business press for facts, not conclusions; SEBI orders for what to avoid; other investors' letters and memos for frameworks (never for tips).
Idea journals: write down every idea with its source and the date; review quarterly which sources produced positions and which produced kills.

5. Circle of competence¶

Buffett's phrase; the practical version: the set of businesses whose economics, KPIs and risks you can explain to a sceptic without notes, and for which you know what would prove you wrong. Draw it explicitly:

Inside (for a vol trader turned fundamental investor, illustrative)	Edge	Outside (for now)
Exchanges, brokers, AMCs, depositories	You understand the plumbing and the regulatory calculus from the inside	Pharma (regulatory science)
Lenders with disclosed unit economics	Carry-trade intuition; RoA tree	Real-estate NAV (land valuation judgement)
Industrials with clear KPIs (pumps, components)	Numerical; the course's running example	Biotech, defence electronics (technology judgement)
IT services (AI transition as a vol event)	Regime thinking	Media content

The circle grows by doing case studies (Module 13) and capstones (Module 15), one sector at a time. Ideas outside it go to the queue with a flag, not to the deep dive.

6. The watchlist and kill criteria¶

Every idea that survives triage but is not bought (price too high, catalyst absent) goes on a watchlist with: the price at which it becomes attractive (from 06.8), the event that would change the view, the date to revisit, and the kill criteria (what would remove it). Kaveri, from this course's analysis: watchlist; attractive below ₹260 with receivable days < 85; revisit at Q3 FY27 results; kill if the pledge exceeds 15% or the auditor qualifies.

Trader's lens

Idea generation is signal research. Screens are factors with known crowding; events are catalysts with known base rates; insider buying is informed flow. The discipline that transfers is the log: record every signal, what you did, and what happened, and let the realised hit rate — not the story — decide which sources you keep.

India notes

Screener.in's fields and their definitions (ROCE, OPM) are on its site; export to Excel for anything beyond a first pass. Tijori, Trendlyne, Tikr and the exchanges' own screeners offer other fields (pledges, promoter holding changes, bulk deals).
Mutual-fund portfolio disclosures are monthly and free; superstar-investor tracking is a cottage industry — the base rate for following it is poor without your own work.
SME-exchange stocks and stocks under ASM/GSM should be excluded from screens by default (09.5).

Common mistakes

Screening on single-year numbers.
Buying the screen output instead of researching it.
Following famous investors' holdings without knowing their thesis, cost or exit.
No kill log — the same bad sources keep producing.
A circle of competence defined by what you own rather than what you understand.

Key terms¶

Term	Meaning
Funnel	The staged filtering of ideas from sources to positions
Screen	A quantitative filter applied to a universe of stocks
Magic formula	Greenblatt's joint ranking on earnings yield and return on capital
Coffee-can investing	Buying consistent compounders and holding for a decade
Net-net	Market cap below net current assets minus all liabilities
Post-earnings-announcement drift	Tendency of prices to continue moving after a surprise
Circle of competence	The businesses you understand well enough to know when you are wrong
Watchlist	Ideas that passed research but await a price or event
Kill criteria / kill log	Pre-set reasons to drop an idea; the record of dropped ideas and why

Check your understanding¶

Write a Screener query for "profitable, cash-generative, low-leverage mid-caps growing 15%+", then list two traps in its output.

Answer

`Market Capitalization > 2000 AND Market Capitalization < 30000 AND Sales growth 3Years > 15 AND Return on capital employed > 18 AND Debt to equity < 0.5 AND Cash from operations last year / Net profit last year > 0.8`. Traps: cyclicals at peak (3-year growth from a trough); companies whose CFO/PAT is good this year but not over three (add a 3-year check from the export); governance not captured (pledges, RPTs).

A stock hits a 52-week low after a promoter pledge invocation. Idea or kill?

Answer

Usually kill on governance grounds unless the business is sound and the promoter's problem is separable — check who now holds the shares, whether the pledge is fully invoked, and the company's own balance sheet. Forced selling can create bargains, but pledge invocations are mostly the start of a story, not the end.

Why is a demerger a better idea source than a screen?

Answer

It creates a structural, temporary mispricing (forced selling by holders who didn't choose the spun entity; no analyst coverage; no index membership) that screens cannot see because the entity has no trading history — fewer competitors, more edge.

Your circle of competence excludes pharma. A friend's pharma idea looks compelling. What do you do?

Answer

Queue it, flagged "outside circle"; if the sector interests you, start the sector work ([08.4](../08-sectors/04-pharma-and-healthcare.md)) and a case study before any position — or size it as a speculative position with a hard max-loss, acknowledging you cannot judge the thesis-breakers.

Go deeper¶

Joel Greenblatt, The Little Book That Beats the Market — the magic formula and its logic.
Saurabh Mukherjea, Coffee Can Investing — the Indian consistency screen and its caveats.
Screener.in's query documentation and the "Export" feature; Tijori's pledge and promoter-change screens.
Warren Buffett, 1996 Berkshire letter — the circle of competence passage.

← Previous: 10.5 Build-along: the Kaveri model · Module index · Next: 11.2 The research process →