A broker is short 20 lots, lot 50, of the one-month Satpura 50 22,000 straddle on 14% implied volatility. The exchange revalues the position for index moves of minus 6% to plus 6% combined with volatility up or down 4 points. Which scenario is worst, and what margin does it imply?

The worst scenario is a 6% rise with volatility up 4 points, a loss of about Rs 8.56 lakh, so the scenario margin is about Rs 8.6 lakh on Rs 7.15 lakh of premium. A 6% fall with volatility up costs Rs 6.36 lakh. A short straddle loses on movement in either direction and on rising volatility, so the margin model charges for both at once.

Think of a landlord taking a deposit. He does not ask what the tenant will probably break; he asks what the worst plausible damage would cost and holds that much. A scenario margin model does the same: it revalues the position under a fixed set of plausible market moves and holds the worst loss as initial margin. For Anjaneri the position is short 1,000 index units of a straddle worth about 715 points, so every point the straddle gains in value is a Rs 1,000 loss for the firm.

Read the grid one axis at a time. Along the middle row, the index does not move and only volatility changes: the straddle gains or loses about Rs 2.0 lakh for 4 points of volatility. Down the middle column, volatility is unchanged and the index moves: at plus or minus 6% one leg of the straddle is 1,320 points in the money, and the loss is about Rs 5.4 lakh for the fall and Rs 7.8 lakh for the rise. The worst cell stacks the two, because a short straddle is short both gamma and vega: it loses when the index travels and loses again when the market pays more for options. In equity indices a sharp fall usually comes with rising volatility, so for the downside the worst cell is also the likely pairing, not a coincidence the model invented.

Notice the asymmetry. The plus 6%, volatility up cell, at Rs 8.56 lakh, is worse than minus 6%, volatility up, at Rs 6.36 lakh. With a 7% rate the strike is discounted: deep in the money, the call at 23,320 is worth at least spot less the present value of the strike, about 1,448 points, while the put at 20,680 is worth about the present value of the strike less spot, 1,192 points. The margin is set by the plus 6% cell, but the risk a desk lives with is the minus 6% cell, because that is the one that arrives with rising volatility.

Three things. The grid stops at 6%; a gap of 10% overnight would cost far more, roughly Rs 14 lakh if volatility jumped to 20%, and the margin does not pretend to cover it. The model revalues instantly, so it gives the short straddle no credit for a day of time decay, which is conservative in calm markets. And the exchange adds other charges on top of the scenario loss, such as an exposure margin and minimum charges for short options, whose current values must be read from the exchange's own circulars. A firm sizing its book on the grid's worst cell is sizing on a regulator's plausible case, not on its own worst case, and its internal limit should sit beyond it.