I tested 108 versions of a drawdown rule

Most "smart SIP" strategies follow the same idea: keep investing every month, but when the market falls sharply, move some money to safety and buy back in once things settle. I wanted to know whether this works, under which rules, and by how much.

The setup

Every month, ₹1 lakh goes into the index. If the index falls more than a set percentage from its peak, part of the holding moves to a liquid fund earning 6%. The money goes back into the index either when it recovers to its previous high, or when the fall passes a deeper threshold that marks it as a crash worth buying into.

I tested six sell triggers (10% to 35%), six buy-back triggers (25% to 50%) and three amounts to move (30%, 50%, 70%). That is 108 combinations, each a full 257-month simulation on the same index data: NSE Indices’ official month-end closes for Nifty Midcap150 Momentum 50, April 2005 to August 2026.

What won

Selling at a 15% fall, buying back once the fall passed 25%, and moving 70% of the holding each time returned 22.20% a year, against 20.97% for plain monthly investing. On ₹2.57 crore invested over twenty years, that gap is worth about ₹6.7 crore.

Its neighbours were not so lucky. Only 19 of the 108 combinations beat plain investing, and the 20%/35% pair that won when I first ran this lost at all three amounts moved.

Why I don't fully trust it

The gain comes from a handful of trades. The winner acted seven times in 21 years, three of them in 2008, and has sat 70% in a liquid fund since February 2025. A rule that acts seven times in twenty years has not been tested many times.

Most combinations lost. Only 6 of 36 beat plain investing at the middle setting (50% moved). If you had picked thresholds without hindsight, the odds were against you.

Small changes matter a lot. A reliable rule usually sits among neighbours with similar results. Here, changing one setting by a step could cost more than a percentage point a year.

The data decided the winner. I first ran this on an index series read from a chart image. It said 20% and 35% won, by 1.6 points a year. On NSE’s official closes, used above since September 2026, that pair loses. Near a threshold, a small difference in the data shifts a trade by a month, and the whole strategy depends on threshold crossings.

Costs are not included. There are no exit loads and no short-term capital gains tax. The winning version made a single move of ₹26.1 crore. With tax, the result would look different.

What I take from it

Not "use 15% and 25%", and certainly not the 20% and 35% that won on the chart-read data. A disciplined rule for reducing risk can beat doing nothing, but the benefit is smaller and less reliable than the headline number suggests, and most versions of the idea do not beat investing every month. The most useful thing to do is change the thresholds yourself in the Strategy Lab and see how quickly the advantage disappears.

Nothing here is investment advice. See the disclaimer.