PE zone switching adds ₹10.27 lakh. The moving average ratio adds ₹3.07 lakh. Continuous PE percentile allocation loses up to ₹13.88 lakh. The same signal, three implementations, three very different outcomes.

The premise

PE ratio is the most widely cited valuation metric for Nifty. At extremes — when PE is far above or below its long-run average — it carries genuine predictive information about forward returns. The question this group of strategies tests is whether that information can be operationalised into a deployment rule that outperforms plain SIP over 30 years. Three implementations are tested, ranging from simple zone switching to continuous fractional allocation based on PE percentile rank.

How all three strategies work

All three use the same bank framework: ₹10,000 deposited monthly, compounding at 5% annualised. Transaction cost 0.03%. Units held permanently. Benchmark is a capital-matched monthly SIP. SIP XIRR: 11.9907% for Strategies 10 and 11, 11.9724% for Strategy 16.

Strategy 10 - PE Zone SIP: The strategy computes a rolling PE moving average over a lookback window (3M, 6M, 12M, 24M, or 36M). If the current PE is below the moving average, the zone is "cheap" and the full bank balance deploys into Nifty. If PE is above the moving average, the zone is "expensive" and the bank accumulates. Five variants tested — one per lookback window.

Strategy 11 - PE MA Ratio: The strategy computes the ratio of current PE to its moving average over a lookback window (24M, 36M, or 48M). If the ratio is below 1 (PE below its own average), the full bank deploys. Above 1, the bank accumulates. Three variants tested.

Strategy 16 - PE Percentile Continuous Allocation: Each month, the strategy computes the PE's percentile rank within a historical lookback window (36M, 60M, 120M, or ALL). A sensitivity parameter (0.6, 0.8, 1.0, or 1.2) and a deployment floor (10%, 20%, or 30%) control how aggressively the fraction scales with the percentile. High percentile means expensive, so the deployment fraction falls. Low percentile means cheap, so more deploys. 48 variants tested.

Strategy 10: all five outperform

Every PE zone variant beats SIP. The best is the 36M lookback: XIRR 12.1434% versus SIP's 11.9907%. Alpha +0.153%. Additional wealth over SIP: ₹10.27 lakh. 207 trades executed. Idle bank at end: ₹1.27 lakh.

The 6M lookback is second: XIRR 12.1398%, alpha +0.149%, additional wealth ₹10.02 lakh.

The 12M lookback adds ₹8.67 lakh. The 24M adds ₹7.74 lakh. The 3M lookback, the noisiest signal, still beats SIP by ₹3.00 lakh with alpha of +0.045%.

The performance ordering is almost perfectly monotonic with lookback length — longer lookbacks outperform shorter ones. The 36M moving average is slow to react, which means it stays in "cheap" zone during sustained rallies and deploys aggressively during prolonged cheap periods. The 3M average reacts quickly, firing more frequently but with less conviction, closer to always-invested SIP behaviour.

All five variants deploy between 185 and 209 times across 367 months. The strategy is not trying to time every month — it simply deploys all accumulated capital when PE drops below its own trend and waits when PE is elevated. On Nifty, where the PE has spent long stretches above its historical average during bull runs and compressed sharply during corrections, this binary rule catches enough of the discount windows to generate real alpha.

Strategy 11: all three outperform

The PE MA ratio sweep produces three outperforming variants. Best is the 48M lookback: XIRR 12.0369%, alpha +0.046%, additional wealth ₹3.07 lakh. 326 trades.

The 36M lookback adds ₹2.07 lakh. The 24M adds ₹1.26 lakh.

The results are real but smaller than Strategy 10. The difference is structural. Strategy 11 deploys based on the same PE vs moving average signal, but it fires more often — 326 to 342 trades versus Strategy 10's 185 to 209. The ratio signal is more sensitive because it reacts to the PE crossing its own average in either direction rather than switching zones. More frequent deployment means the strategy behaves more like SIP, which compresses the alpha. The signal is correct but the implementation blurs it.

Strategy 16: zero for 48

Every single PE percentile continuous allocation variant underperforms SIP. The best result is the 120M lookback with sensitivity 0.6 and any floor — XIRR 11.8519%, alpha -0.121%, trailing SIP by ₹7.86 lakh. The worst, a 60M lookback with sensitivity 1.0 and 10% floor, trails by ₹13.88 lakh.

All 48 variants execute exactly 324 trades out of 367 months. The floor parameter makes no difference to the results — identical XIRR across all three floor values for each lookback and sensitivity combination. The floor is never binding enough to change the deployment meaningfully.

The problem is the same one that ran through Strategy 18 in the previous post. Continuous fractional allocation means the strategy always deploys something, but it always holds something back too. When PE is at the 80th percentile, the strategy deploys 20% to 30% of the bank. When PE is at the 20th percentile, it deploys 70% to 80%. The remaining balance earns 5% in the bank while Nifty compounds at 12%. The fractional drag accumulates over 30 years.

Higher sensitivity (1.2) makes things worse because it sharpens the scaling, holding back more in expensive months and deploying more in cheap months. But the cheap months are often also high-volatility months where Nifty is recovering fast — deploying a larger fraction there helps, but not enough to overcome the drag from the high-PE restraint across bull market years.

Longer lookbacks marginally outperform shorter ones, mirroring the finding from Strategies 10 and 11. But even the best configuration is stuck at -0.121% alpha.

Reading all three together

The contrast is stark and the lesson is specific. Binary deployment — all in when cheap, hold when expensive — works on Nifty's PE signal. Continuous fractional deployment on the same signal does not.

Strategy 10 with a 36M lookback deploys 207 times in 367 months. When it deploys, it deploys everything. When it waits, it waits with the full bank compounding at 5%. The signal is sharp enough that the binary decision outperforms continuous partial allocation by a wide margin. The 36M lookback is slow enough to filter noise and fast enough to catch the major cheap windows 2003, 2009, 2013, 2020.

Strategy 16 never fully commits. It is always partially in and partially out, which means it never gets the full benefit of a cheap window and always carries the cost of holding back capital during expensive windows.

The deeper point: valuation timing on Nifty works when you act on it decisively. PE-based signals have genuine information content, but that information is best used as a binary switch, not as a continuous dial.

Verdict

Strategy 10 goes 5 for 5, adding between ₹3 lakh and ₹10.27 lakh over SIP depending on lookback. Strategy 11 goes 3 for 3, adding between ₹1.26 lakh and ₹3.07 lakh. Strategy 16 goes 0 for 48, losing between ₹7.86 lakh and ₹13.88 lakh. All three use the same underlying signal. The implementation decides the outcome.