The central bank hikes 100 basis points when 25 was expected. Two-year yields rise 80 basis points, ten-year yields 30, the currency 3% and equities fall 6%. How would a macro fund have expressed the view in advance, and how would it size the trade on Rs 500 crore at an 8% volatility target?

Express it as a DV01-matched 2s10s flattener, sized to about 4% volatility: Rs 41.8 lakh per basis point of spread. That is about Rs 2,235 crore paying two-year and Rs 606 crore receiving ten-year. The spread narrowed 50 bp, making Rs 20.9 crore, 4.2% of NAV, more per unit of risk than paying two-year (3.6%), the currency (1.7%) or the equity short (1.3%). An equal-notional flattener would have lost Rs 5.7 crore.

If you expect one shop on a street to raise its prices, the clean bet is that shop against the street, not the city's whole shopping bill, which moves for a hundred other reasons. The view here is narrow: the bank will tighten more than the 25 basis points priced. The two-year yield is mostly the market's guess at the policy rate over the next two years, so it takes the surprise first. The ten-year yield also carries long-run growth and inflation, and a bank that tightens hard lowers both, so it rises less. A hawkish surprise therefore flattens the curve, and a flattener isolates that one effect while hedging out the general level of yields. Equities and the currency respond too, but through earnings, flows and risk appetite as well as the policy rate.

Basis points and percentages cannot be compared directly, so convert every move into how many years of normal volatility it represents. The equity fall of 6% is only 0.33 of a year's 18% volatility: a big headline, a small signal. The currency's 3% is 0.43. Paying two-year rates captured 0.89. The flattener's 50 bp is measured against the spread's own volatility, which is far lower than either yield's because the two yields mostly move together.

The fund's 8% target on Rs 500 crore is Rs 40 crore of annual volatility. With about four independent ideas at equal risk, each can run at 8% divided by the square root of four, so the volatilities add back to the target. One idea gets 4%, Rs 20 crore a year, and dividing by the spread's 47.9 bp volatility gives Rs 41.8 lakh of P&L per basis point of spread. Each leg must carry that DV01: two-year swaps at 1.87 duration move Rs 1.87 lakh per bp per Rs 100 crore, so the leg is Rs 2,235 crore; ten-year swaps at 6.9 need only Rs 606 crore. Gross notional of Rs 2,841 crore looks large against Rs 500 crore, but notional is not risk; the DV01 is.

Because a ten-year swap moves 3.7 times as much per basis point as a two-year swap. Pay Rs 1,000 crore of each and the two-year leg gains Rs 15.0 crore on its 80 bp, while the ten-year leg loses Rs 20.7 crore on its 30 bp. The equal-notional trade is net long Rs 50 lakh of duration per basis point, so it is a bet on falling yields dressed up as a flattener, and it lost Rs 5.7 crore on the day the view came true.

Suppose instead that global bond markets rally and both yields fall 40 bp for reasons unrelated to the bank. The two-year payer loses Rs 8.9 crore; the matched flattener loses nothing, because a parallel move leaves the spread alone. That is the hedge the trade buys. It does not make the view safe: if the bank holds when 25 was priced, and two-year yields fall 20 bp against 5 for the ten-year, the flattener loses Rs 6.3 crore, 1.25% of NAV. The limitations are real. Correlations jump around policy days, so the hedge ratio set from normal volatility can drift; carry and roll-down cost or earn a few basis points a month and matter if the surprise is delayed; and a surprise that is partly priced before the meeting leaves less to collect. Close or cut the position once the market prices the bank's new path, because the edge was the gap, not the hike.