Largest compounders
The fastest growers among those that grew in at least three of the five years before the window.
₹20L put in on 1 Aug 2026 and left alone until 1 Sept 2026. Whole shares, dividends counted, nothing rebalanced or sold.
from ₹20L
over 1 month, not annualised
of the 20 the rule wanted · 15 lost money
₹20L, month by month
20 holdings plus ₹40,914 in cashThe line is the whole position — shares at each month-end, plus dividends as they were paid, plus the cash that never got invested. It is indexed to 100 at the start, so the shape is the return.
The screen, and when it ran
Strictly before the window it is measured overEvery company in the universe was scored on how many of the 5 years ending 1 Aug 2026 its share price rose in. 155 cleared the bar of 3 of 5, across 17 sectors. 18 qualified but were left out because their sector had already contributed its maximum of 3 — that cap is the only thing keeping this pack spread, and it is doing real work.
Not one day of the screen overlaps the window being measured. Screening and measuring over the same years would pick the companies that rose and then report that they rose, which is arithmetic rather than evidence. Everything below is out of sample.
What the rule could actually do with the money
The part most tools skip₹19.6 L bought shares. ₹40,914 could not buy a whole share of anything the rule wanted and sat in cash for the whole 1 month.
Return on the money that was invested: -3.8%. The headline above is lower because it counts the idle cash, which is what would really have happened.
| Large cap | 5 | 24% |
| Mid cap | 6 | 31% |
| Small cap | 9 | 45% |
This pack imposes no size split. The mix above is simply what the screen returned — worth seeing, but it was not aimed at and it is not a miss.
points ahead of the whole-market pack at the same amount over the same window (-4.6%).
Where the return came from
Price movement and dividends, kept apartWhat the shares themselves were worth at the end, less what they cost.
3 of 20 holdings paid something across the 1 month. Dividends are counted as cash when paid and are not reinvested, so nothing here compounds.
This pack finished down, so there is no gain to take a share of. The ₹1,448 of dividends reduced the loss; the price change did the rest.
The two figures add to the gain exactly — every rupee of return is either a price move or a dividend — and the split matters because they do not behave alike: a dividend is money received, a price gain is only realised if it is sold. The share is not a yield, and a smaller share is not a smaller dividend. It shrinks when prices run, because the denominator grows: across this universe income is about a sixth of the one-year gain and about a twentieth of the five-year gain, from much the same dividends. In a window where prices fell, income can be the only part that was positive.
Sign in to see which companies the rule actually held, and what each one did
Guests see a sample. Members get every company, five-year fundamentals, classified developments, a watchlist and the decision journal.
Sign in with GoogleWhy these sectors
The grouping is a judgement — here is the one that was madeLargest rather than steadiest. The consistency requirement is kept low enough to be a filter rather than the selector, and the ranking is then pure compound rate. Read it against Unbroken growth: the two answer different questions about the same universe, and on the period tested they do not agree.
Coverage. 200 of 200 companies in this theme had month-end prices at both ends of the window; the rest were left out rather than filled in. Every ranking used a share count the market had already seen at the start of the window.
Target split 10/6/4 companies · equal weight inside each size band · prices are month-end closes from the adjusted series, so share counts are on today’s basis.