It takes about twenty minutes to build a ten-year table for an Indian company from a free data provider: revenue, operating profit, profit after tax, equity, debt, cash flow from operations, ten rows deep. The table looks authoritative. It has decimal places. You will draw a compound growth rate from it, an average margin, a return on capital across a cycle, and each of those numbers will be quoted in your own notes for years afterwards as though it were a measurement.
The four preceding lessons each described one way that table can be quietly wrong. This one is the procedure that catches them, and it is deliberately unambitious: the aim is not a perfect decade but an honest one, where the breaks are marked and the averages are taken only across stretches that belong together.
A family marks their children’s heights on a door frame every birthday. Ten years of marks, and then the family moves house and the new frame starts again from nothing. Somebody helpfully copies the old marks across, measuring from the new floor — which is two inches higher, because the new house has tiles the old one did not. The marks are all genuine, the transcription was careful, and the resulting growth chart shows a child shrinking in the year they moved.
A ten-year financial series is the same door frame. Each individual year was measured properly. What breaks it is the joins — a division sold, an entity merged, a standard adopted, a life reassessed — and the joins are exactly what a data provider is worst at recording.
The six breaks, and the tell for each
| The break | The tell | Where to look |
|---|---|---|
| A division sold or demerged | A prior-year column smaller than the one published a year earlier; a "discontinued operations" line | The disposal note, which gives both bases for both years |
| A scheme with an appointed date | A prior-year column larger than the one published a year earlier; a jump in share count with no placement | The scheme note, and the scheme documents filed with the exchanges |
| An acquisition from a third party | Revenue jumps, comparative untouched, goodwill appears or grows | The business combinations note, which gives the acquisition date and often the part-year contribution |
| A new accounting standard | Every company in the industry steps in the same year; a transition note appears | The transition note, which says whether comparatives were restated or not |
| A change in estimate | A margin or depreciation ratio steps once and stays; no restatement anywhere | The accounting policies note and the fixed asset note, which disclose the effect on the current period |
| A change in share count | Per-share figures step while absolute figures do not | The share capital note, and the corporate actions history — bonus, split, rights, conversion, buyback |
- 1Start from the reports, not the table
Open the most recent annual report and the one from three years ago. You are not reading them yet — you are checking whether the older report’s current-year column matches the newer report’s three-year-old column. Where they match, the basis has held. Where they do not, you have found a break in under two minutes, and it will be one of the six above.
- 2Pick a base year and hold it
Choose the earliest year you are willing to defend as being on the same basis as today, and start the series there. A clean seven years beats a contaminated eleven, and the reason to write the base year down is that it stops you quietly extending the series backwards later when you want a longer track record.
- 3Repair the per-share column first
Bonus issues and splits are mechanical — every historical per-share figure divides by the same factor, and applying it is a minute of work. Rights issues are not mechanical, because a rights issue priced below the market contains an element of bonus, and the historical adjustment needs a factor rather than a simple divisor. Data providers publish adjusted series for exactly this reason; if you are adjusting by hand, adjust for the mechanical events and note the rights issues rather than guessing at them.
- 4Repair the entity, or admit you cannot
For a disposal or a common-control merger, the notes give you enough to restate two or three years by hand. Beyond that the disclosure runs out, and restating a decade from a single year’s note is guesswork wearing a spreadsheet. Where you cannot repair, split the series in two and say so.
- 5Take averages only within a stretch
This is where the discipline pays. A five-year average margin, a ten-year return on capital and a compound growth rate are all averages, and an average that crosses a break is a statement about two different companies. Compute them inside each stretch, compare the stretches by eye, and resist the urge to produce one number.
- 6Write the changelog on the sheet itself
One line against each affected row: what changed, in which direction, and where you found it. This is the step everybody skips and the only one that still helps you in three years, when you have forgotten every detail of the analysis and the chart is all that is left.
- Ten rows, all from one data provider, all to two decimal places
- One compound growth rate covering the whole period
- A ten-year average margin quoted in the thesis
- A return on capital series with an unexplained step in year six
- No record of where any figure came from
- Seven rows on one basis, three on another, with the join marked
- Two growth rates, one for each stretch, and a sentence on what happened between them
- Averages computed inside each stretch and compared by eye
- The step in year six labelled: modified retrospective adoption, comparatives not restated
- A changelog line against every affected row naming the note it came from
You are building a ten-year series and discover that a division was demerged in year six and a promoter-owned company merged in during year eight. You can restate two years around each event from the notes, but not the whole decade. What is the most defensible thing to do?
Module checkpoint: the base that moved
5 questions. Answers are revealed once you submit all of them.
1.A company sold a major division this year. Its report shows revenue of ₹6,200 crore against a prior year of ₹5,900 crore. Your own record of last year’s published revenue is ₹8,400 crore. Which comparison is meaningful, and what is the ₹2,500 crore difference?
2.Why is the March quarter of an Indian listed company systematically lumpier than the three quarters before it?
3.Two companies each announce that they have merged another business into themselves. Company A’s prior-year comparative rises from ₹2,100 crore to ₹2,700 crore in the new report; Company B’s prior-year comparative is unchanged while goodwill appears on its balance sheet. What distinguishes the two transactions?
4.A company cut its annual depreciation charge from ₹200 crore to ₹120 crore by reassessing the remaining useful life of plant carried at ₹3,000 crore from fifteen years to twenty-five. Its profit after tax rose from ₹340 crore to ₹400 crore. Which statement is correct?
5.You are computing a ten-year average operating margin. Years one to five are on one accounting basis, and years six to ten follow a transition where comparatives were not restated. What is the correct treatment?
Har janamdin pe bachchon ki lambai darwaze pe lagti thi. Ghar badla, naye darwaze pe purane nishaan utaar diye — par naye ghar mein tile lag gaye the, farsh do inch ooncha. Nishaan sab asli, utaarna bhi dhyan se, aur chart mein bachcha us saal chhota ho gaya. Das saal ka table bhi aisa hi hai: har saal sahi naapa gaya, jod galat hai. Purani annual report ka column nayi report se milao — jahan farak mile, wahin nishaan lagao aur average sirf ek tukde ke andar nikalo.
- Compare an old annual report’s current-year column with a new one’s historical column — where they differ, you have found a break in two minutes.
- Pick a base year you can defend and write it down, so the series does not quietly grow backwards later.
- Repair per-share figures for bonus issues and splits mechanically; note rights issues rather than guessing at the adjustment factor.
- Take averages and growth rates only inside a stretch that shares one basis, never across a break.
- Where a break cannot be repaired, fall back on a coarser continuous measure — cash from operations, total assets, dividends paid — and label the events on the chart.
Mark it done to track your progress through the curriculum.
Common questions
Short, direct answers to what people ask about this topic.
- how to check whether a ten year table from a screener is comparable
- Open the latest annual report and one from about three years ago, and check whether the older report’s current-year column matches the newer report’s three-year-old column. Where they match, the basis has held; where they do not, you have found a break, and it will be one of six — a disposal, a scheme with an appointed date, a third-party acquisition, a new accounting standard, a change in estimate or a change in share count. The check takes about two minutes and belongs before any ratio is drawn.
- base year meaning when building a long financial series
- The base year is the earliest year you are willing to defend as being on the same basis as today, and the series starts there rather than as far back as the data goes. Writing it down matters because it stops the series quietly being extended backwards later, when a longer track record looks appealing. A clean seven years is worth more than a contaminated eleven.
- how do I adjust historical EPS for a bonus issue or a stock split
- Bonus issues and splits are mechanical: every historical per-share figure divides by the same factor, so a one-for-one bonus halves every earlier earnings per share and dividend per share. Rights issues are not mechanical, because a rights issue priced below the market contains an element of bonus and needs an adjustment factor rather than a simple divisor. Data providers publish adjusted series for exactly this reason; adjusting by hand, handle the mechanical events and note the rights issues rather than guessing at them.
- can I take a ten year average margin if the company sold a division midway
- An average that crosses a break is a statement about two different companies. Compute the average margin inside each stretch that belongs together, compare the stretches by eye, and resist the urge to produce one number for the decade. The same applies to a ten-year return on capital or a compound growth rate spanning a merger with an appointed date.
- a compound annual growth rate taken across a change in the reporting entity describes
- A change in the reporting entity, not a change in trade. The arithmetic can be perfectly correct while the two endpoints describe different collections of businesses — one before a division was sold or a merger was backdated, one after. Which is why the base of a series is worth checking before any growth rate drawn from it goes into a note you will still be quoting in three years.