Coffee Can Investing (Buy & Forget)

Buy a handful of great compounders, seal the can, and do not touch them for a decade or more.

The thesis — why it works

The name comes from old-time settlers who kept their valuables in a coffee can under the mattress rather than trading them. Robert Kirby's 1984 essay described a client who never sold a single stock her late husband had bought — and, thanks to a few enormous winners left fully uncut, ended up beating professionally "managed" portfolios that kept trimming winners and trading actively.

Saurabh Mukherjea formalized this for India in Coffee Can Investing, with a strict, mechanical entry filter: only companies that grew revenue by ≥10% and earned ROCE ≥15% in every single year, not just on average, over the last decade. Once bought, the portfolio is genuinely never touched — no stop-losses, no profit-booking, no rebalancing — for a minimum of ten years. The edge comes from two things: letting a handful of exceptional compounders do the heavy lifting undisturbed, and eliminating the tax drag, transaction costs and behavioural errors (overtrading, panic-selling) that come from active management.

When to use it

You genuinely have a decade-plus horizon and won't be tempted to interfere.
You want minimum-maintenance investing — pick once, review rarely.
You trust a strict, mechanical entry filter more than your own judgment about when to buy or sell.

When to avoid it

You may need the money, or want to actively manage/rebalance, within the next several years.
You can't resist checking prices and tinkering — the strategy only works if you truly do nothing.
You want valuation-based entries and exits — this method deliberately ignores price/valuation, using only quality-of-business filters.

The screen (Screener.in)

The original test is about year-by-year consistency, which a single screen snapshot can't fully capture — use the query to shortlist, then verify consistency manually.

Criterion	Rule	Why
Market cap	`> ₹100 Cr`	excludes illiquid, barely-listed names
Sales growth (10-yr)	`> 10%`	durable revenue growth over a full decade
Return on capital employed	`> 15%`	efficient, quality use of capital
Debt to equity	`< 0.5`	a decade-long unmonitored hold can't survive a balance-sheet blow-up

Copyable Screener.in query:

Screener.in query

Market Capitalization > 100 AND
Sales growth 10Years > 10 AND
Return on capital employed > 15 AND
Debt to equity < 0.5

Screener's growth and ROCE fields are CAGR/trailing figures, not "every single year" checks. Open each candidate's 10-year data view on Screener and confirm growth stayed ≥10% and ROCE stayed ≥15% in each individual year — that year-on-year consistency, not the average, is the heart of the original test.

The procedure

Run the query for a mechanical first-pass shortlist.
Open each survivor's 10-year financial history and check, year by year, that revenue growth and ROCE cleared the bar every single year. Discard anything with even one bad year.
Sanity-check governance — promoter holding trend, related-party transactions, audit history — since you won't be actively monitoring for a decade.
Build a basket of 15–20 survivors, equal-weighted or lightly conviction-weighted.
Seal the can: no stop-loss, no profit-booking, no rebalancing. Revisit the portfolio only after ten years, or immediately if a genuine fraud/delisting event occurs.

Decision & risk rules

Position size: equal-weight across 15–20 names at initiation — diversification is your only risk control since you won't actively monitor.
Holding period: minimum 10 years; this is the strategy, not a guideline.
Rebalance: none, by design — the only exception is a hard fundamental break (fraud, delisting, business-model collapse).
Entry: stagger purchases over a few months; the entry filter, not price timing, does the risk-control work.
Exit / invalidation: only for a change in the underlying facts — governance fraud or genuine business obsolescence — never for price volatility or "it looks expensive now."

Common mistakes

Checking the portfolio often and getting tempted to trim winners — this kills the very skew (a few huge winners) the strategy depends on.
Relying on the average/CAGR growth and ROCE figures instead of verifying genuine year-by-year consistency.
Ignoring governance red flags because "I'm not touching it anyway" — a decade-long unmonitored hold makes you more exposed to fraud, not less.
Applying the filter loosely to companies you merely like — the strategy only works if the entry bar was genuinely strict.