Twelve-Month Momentum and Draw-Up Thresholds on Nifty: Six Bug Fixes Halved the Edge

Lineage: Mirrors the drawdown-buying study with the signal flipped from weakness to strength, and replaces Strategy 14's published figures with a corrected rebuild.

The premise

The dip-buying studies flipped the sign and found nothing usable. This study flips it back and asks whether confirmed strength is a better deployment trigger than confirmed weakness, which is a more defensible starting point: cross-sectional momentum has decades of academic support, while buying falls does not.

Four signals test the question at different frequencies. Strategy 14 checks the trailing twelve-month return excluding the most recent month, the standard twelve-minus-one construction from the momentum literature. Strategies 7 and 9 measure how far the index has climbed past a threshold, at weekly and monthly resolution respectively. Strategy 13 changes the reference point entirely and watches how close the index sits to its own all-time high.

A 7.5% monthly gain is rare and carries information. A 1% weekly gain happens constantly and carries almost none. Strategy 13, meanwhile, is not measuring momentum at all: proximity to a price peak is closer to a valuation reading wearing a momentum label, and it is included here precisely to show what happens when the label and the mechanism diverge.

How the strategies work

Strategy 14 sweeps nine thresholds from −20% to +20% in 5% steps, rather than the single binary rule the earlier write-up described. A 39-point grid from 1.0% to 20.0% in 0.5% steps runs at weekly resolution as Strategy 7 and at monthly resolution as Strategy 9. Strategy 13 sweeps fifteen thresholds from −7% to +7% in 1% steps, taking at most one trade a month at the next day's open.

That gives 102 combinations rather than the 87 an earlier draft counted, because the surviving workbooks carry a wider grid than the published article used and because Strategy 14's binary framing was incorrect.

The setup

All four share one bank framework: ₹10,000 deposited on the first trading day of each month, compounding at 5% annualised until a signal releases it, a 0.03% transaction cost per trade, and every unit held permanently. No capital gains tax applies, because no gain is ever realised. The benchmark is a capital-matched monthly SIP on the same schedule and cost: 11.9907% XIRR, ₹3.2641 crore final on ₹36.7 lakh contributed.

Maximum drawdown and Calmar appear in none of these four SUMMARY sheets, so this study works from XIRR and final wealth alone.

The Strategy 14 figures here come from a rebuild that corrected six measurement defects, and the largest of them matters enough to state plainly. Its XIRR was computed on deployed cash flows while the benchmark's was computed on contributed cash flows, so bank interest never counted against the strategy while still counting toward its final wealth. A second defect compounded the bank monthly instead of daily. Correcting both, along with three smaller ones, roughly halves the reported edge. Sharpe and Sortino rise sharply as well, from 0.0613 to 0.8243 and from 0.1184 to 1.5885, because the earlier formula was not annualised rather than because risk fell.

The best combinations

The rebuilt Strategy 14 at a 0% threshold produces ₹3.3034 crore against the benchmark's ₹3.2641 crore across 269 executed trades. Strategy 9's best setting is behind it by roughly ₹1,550 across 51 trades, which is less than a single month's contribution and is not a meaningful distinction.

Fourteen of Strategy 9's 39 thresholds beat the benchmark, clustered between 3% and 10%. Strategy 7's best setting, a 1% weekly threshold, still loses by ₹21,197. Strategy 13's best setting, a −7% threshold, trails by ₹42.89 lakh.

What the two leaders share is patience. Strategy 14 sat out 97 of roughly 366 signal months, holding through long stretches without a trade rather than deploying on every positive print.

What the key parameters do

Frequency rather than threshold size decides this study. Twelve-month and monthly momentum work; weekly and all-time-high proximity do not. A 7.5% monthly move carries information about what follows it, while a 1% weekly move is ordinary variation.

Strategy 9's outperforming band runs from 3% to 10% and falls away on both sides. Below 2.5% it behaves like a noisy SIP, deploying too readily to be selecting anything. Above 18% it fires once or not at all.

Weekly draw-up deteriorates monotonically as its threshold rises, from an almost-benchmark 11.9875% at 1% down to pure bank compounding at the top of the sweep, where the highest thresholds never trigger a single trade and the money simply earns 5% while the index compounds far ahead of it.

All-time-high proximity is disqualified at every one of its fifteen levels, and the reason is worth separating from the others. It measures distance from a price peak, and Nifty sits near its peak precisely when valuations are stretched rather than when a fresh uptrend has just confirmed. The signal fires at the wrong moments by construction.

Where it fails

Strategy 7 fails at all 39 thresholds because weekly moves carry no forward information. A weekly spike is usually a single-session event followed by reversion, so buying at the next open means buying the top of a move that has already finished.

Strategy 13 fails at all fifteen levels for the opposite reason: it accumulates cash during corrections, which is when the best entries exist, and deploys at fresh peaks, which is when they do not.

Both leaders carry a softer weakness that the table conceals. Strategy 9's 7.5% optimum is one point on a 39-value sweep, found empirically rather than derived, and nothing in this study tests whether it holds in the second half of the window. Strategy 14's rule has decades of published support behind it and only one threshold that matters, which makes it the more durable of the two even though its rebuilt edge is now the smaller number.

The rebuild itself is the broader caution. Six defects in one script moved a headline figure by roughly half, and the direction of the correction was not obvious in advance. Any figure in this series that has not been through the same reconciliation should be read with that in mind.

Verdict

After the rebuild, Strategy 14 and Strategy 9's best setting are functionally tied at ₹3.92 lakh against ₹3.91 lakh, a gap of about ₹1,550 across thirty years. The published claim of a clear ₹4.34 lakh win was a product of the measurement defects rather than a real margin, and correcting it is the most useful thing this study does. Weekly draw-up and all-time-high proximity can be set aside entirely: nought winners across 54 combined thresholds. Choosing between the two survivors is a choice between one non-fitted rule and one empirically tuned threshold, and on a payoff this close, the version with fewer degrees of freedom is the safer bet out of sample.

Appendix: assumptions and methodology

• Data. Nifty at index level. The surviving output does not state whether this is Nifty 50 or Nifty 500, and that is carried as a gap.
• Benchmark. Capital-matched monthly SIP, ₹10,000 on the first trading day of each month, 0.03% cost: 11.9907% XIRR, ₹3.2641 crore final. Maximum drawdown and Calmar do not appear in the SUMMARY sheets.
• Tax. None modelled, and none applies. Units are bought and held permanently, so no gain is ever realised.
• Signal definitions. Strategy 14 uses the trailing twelve-month return excluding the most recent month, against thresholds from −20% to +20% in 5% steps. Strategy 7 uses weekly close-to-close draw-up from 1.0% to 20.0% in 0.5% steps, and Strategy 9 the same grid monthly. Strategy 13 uses distance from the all-time high, −7% to +7% in 1% steps, at most one trade a month at the next day's open.
• Grid. 102 combinations: 9 plus 39 plus 39 plus 15.
• Execution. Each signal is checked on its own frequency. Idle bank compounds daily at 5% since the rebuild, and every trade costs 0.03%.
• Metrics. XIRR and additional wealth against the SIP. Sharpe and Sortino are reported for Strategy 14 only, now measured as excess over the risk-free rate and annualised by the square root of 252.
• Not modelled. Brokerage beyond 0.03%, securities transaction tax, stamp duty, slippage, dividend reinvestment, year-by-year returns, split-period figures.
• Superseded figures. Strategy 14's published ₹4.34 lakh lead, its 0.0613 Sharpe and its 0.1184 Sortino all predate the rebuild and are quoted above only for comparison.
• Withheld. Nothing. Every threshold and the full grid are stated above, because the finding concerns signal frequency rather than any particular setting.