A metals company's EBITDA over eight years was Rs 300, 450, 600, 250, 150, 400, 700 and 500 crore, and this year's is Rs 500 crore. Peers trade at 6x mid-cycle EBITDA. What EBITDA would you capitalise, what value does that give, and why not use this year's figure?

Capitalise mid-cycle EBITDA of about Rs 419 crore, the eight-year average, which at 6x gives Rs 2,512 crore. This year's Rs 500 crore would give Rs 3,000 crore, 19% more, because 500 is a good year in a business that has earned anything from Rs 150 crore to Rs 700 crore on the same plant. The 6x is a multiple of the average; put it on a peak year and you have paid twice for the cycle.

Because this year is a point on a cycle, and the business is the whole cycle. Eight years on the same capacity produced between Rs 150 crore and Rs 700 crore of EBITDA, so no single year describes what the plant earns; the average, Rs 419 crore, comes closest. A farmer does not value his land on the harvest after a perfect monsoon, and a buyer does not pay for it on the drought year; both look at ten years of yields. Tamrav's Rs 500 crore is 19% above the average and the second-best figure in nine years. The variation is large, a coefficient of variation of about 41%, which is what cyclical means in numbers.

The market applies a steady multiple to a steady number, so the multiple on any given year's EBITDA swings the other way: Rs 2,512 crore is 5.0x this year's 500, 3.6x the peak 700 and 16.8x the trough 150. That is why cyclical stocks look cheapest at the top of the cycle, when earnings are high and the multiple is low, and most expensive at the bottom. A buyer who sees 6x peers and 500 of EBITDA and pays Rs 3,000 crore has applied a mid-cycle multiple to a near-peak year, which counts the good part of the cycle twice: once in the multiple, which already assumes some good years, and once in the base.

When something about the plant or the market has changed. If capacity had grown, older years would need scaling up; if the cost base had fallen permanently, the average understates the future; if the last five years are a better guide to the next five, the Rs 400 crore five-year average is defensible. Including this year in a nine-year average gives Rs 428 crore, and it belongs if the year is nearly complete. What does not belong is an argument that this cycle is different because prices are high now; every peak has that argument attached. Check too whether the 6x is really a mid-cycle multipleA valuation multiple that analysts apply to a normalised, average-over-the-cycle earnings figure rather than to the current year, so that it stays roughly stable as earnings swing.: if peers are priced on this year's earnings at 6x and this year is also their good year, the multiple is already a trough multiple on peak earnings, and applying it to an average would understate.

Close with the view and its limit. Rs 2,512 crore, with a range from the five-year average to the nine-year one, Rs 2,400 crore to Rs 2,567 crore, and a clear note that this year's figure is not a base. The limit of any average is that it assumes the next cycle looks like the last; a plant facing a structural change in its input cost or a new competitor's capacity has a different mid-cycle, and the right number is a forecast, not a history. Say which you are using, because a bid at Rs 3,000 crore is a bet that the next eight years average 500, and nothing in the record supports it.