Last year's best factor is a bad forecast

A friend sent me last year's factor leaderboard this week. Momentum on top, value scraping the bottom. He wanted to rotate his SIP into the winner. Reasonable instinct. It is also the single most common way people lose to a portfolio that does nothing clever.

I asked him to do one thing first. Hold last year's leaderboard up against the year before it. The top tile had moved. It almost always has.

So I ran the trade he was about to make. Every year, buy whatever factor led the year before, hold it, repeat. Over five years that compounds to roughly +58 percent. Then I ran the lazy version. Hold all five factors, equal weight, rebalance on a schedule, never look at the leaderboard. +72 percent. Same five factors. On a 10 lakh book that is about 1.4 lakh of compounding handed away for the privilege of feeling decisive.

The reason is in the rotation. Across those years, last year's number one repeated as this year's number one exactly once in five tries. Momentum led one year and sat dead last two years later. Value bottomed, led, then bottomed again. Trailing return is a baton in the middle of a handoff. Chase it and you grab it right as it leaves the hand.

The blend wins because the rebalance quietly does what you cannot make yourself do. It trims the runner and buys the laggard, on a calendar, with no opinion attached. Cliff Asness called factor timing a siren song back in 2016, and the trailing-return version is the loudest verse.

If you genuinely want to tilt, tilt on something slower than last year. How cheap a factor sits against its own history. That signal ages in years. A hot twelve months is already spent.

The opening's two paths, in rupees

Put the two results on the Rs 10 lakh book. Chasing last year's leader turned it into about Rs 15.8 lakh, a yearly rate of about 9.6 percent. The equal weight blend turned it into about Rs 17.2 lakh, about 11.5 percent a year. Two points a year sounds small. Over five years it is Rs 1.4 lakh, and over twenty it would be far more.

The leader depends on the window you pick

The leaderboard my friend sent was one window. Change the window and the board reshuffles. Here are NSE's factor indices on three windows ending 9 Oct 2026.

Read down each column. Over six months the Nifty Alpha 50 led and value was last. Over two years value led and the Momentum 30 was last. Over five years to 30 Sep 2026 on NSE's factsheets the order changed again, with value first at 22.9 percent a year on total return and quality last at 6.4 percent. Any of these is a true leaderboard. None of them tells you the next one.

Eight quarters of handoffs

Here is the same rotation, quarter by quarter, on four of those indices: momentum, value, quality and low volatility. The leader changed five times in eight quarters.

1. Jan 2025Low Vol 30 led, down 6.6%

   Every factor fell; the calmest fell least. Momentum lost 15.3%.

2. Apr 2025Value 30 led, down 2.8%

   Momentum lost another 16.9%.

3. Jul 2025Momentum 30 led, up 20.2%

   Last quarter's worst became this quarter's best.

4. Oct 2025Value 30 led, up 2.4%

   Momentum slipped 1.5%.

5. Jan 2026 to Apr 2026Value 30 led twice in a row

   Up 7.5% and then 4.0% while quality fell 8.1%.

6. Jul 2026Low Vol 30 led, up 3.0%

   Value dipped 1.4%.

7. Oct 2026Momentum 30 led, down 3.6%

   Least bad in a falling quarter; value fell 7.1%.

The honest part

I ran the chase on these eight quarters too: hold last quarter's leader for the next quarter, against an equal blend of all four put back to a quarter each every quarter. Last quarter's leader repeated in 2 of 7 tries. Yet this time the chase finished ahead, at 96.1 against 88.0 for the blend, rebased to 100, price only and before costs. The reason is one long run: value led four of the eight quarters, and a chaser who happened to be holding it rode most of that.

So the evidence cuts both ways in a short window. Chasing wins when one factor has a long run and loses when leadership flips quickly, as it did in the opening's five years. You only find out which kind of period you were in afterwards, and the chaser also pays tax and costs on every switch, which the blend avoids between rebalances. That is why the free Learn module on factor timing and cycles focuses on slow signals such as valuation spreads, not last year's return. The posts on factor timing and long only factors take it further, and the drift threshold calculator shows when a blend has drifted far enough to rebalance.