What rotation actually describes
Capital does not enter or leave the market uniformly. It concentrates in some sectors and drains from others, and those shifts often begin while the headline index is still flat.
On NSE the observable version is relative strength by sector — how each sector index is performing against the broad market, and whether that is improving or decaying. A sector can be rising in absolute terms while losing relative strength, which is usually the more informative reading.
Why breadth leads the index
An index is a weighted average, so a handful of large constituents can hold it up while most of the market deteriorates underneath. Breadth measures — advance/decline, the share of stocks above their 200-day average, new highs against new lows — show that divergence directly.
Narrowing breadth on a rising index is the classic late-trend condition, and it is invisible if you only watch the index level. The same logic applies per sector: leadership changes show up in a sector's internals before they show in its index.
This is the reason our screening guide puts breadth first in the sequence rather than last.
The three ways rotation studies go wrong
- Survivorship. Testing on today's index constituents silently excludes everything that was removed. The removals are disproportionately the failures, so the result is flattering by construction.
- Look-ahead. Ranking sectors on full-period returns and then "backtesting" that ranking from the start. Nobody could have known the ranking on day one, so the result is unachievable.
- Point-in-time membership. Sector definitions and index membership change. A stock reclassified in 2023 was not in that sector in 2019, and a study that applies today's mapping backwards is testing a portfolio nobody could have held.
These are properties of how a test is built, not knobs to tune afterwards. A rotation result that does not state how it handled all three has not told you whether it means anything.
What a rotation rule has to specify
Vague rotation ideas cannot be tested. A testable one states:
- Universe — which sectors or stocks are eligible, rebuilt at each rebalance rather than fixed today.
- Ranking factor — 3-month momentum, 6-month, risk-adjusted momentum, Mansfield relative strength, proximity to 52-week high.
- How many to hold and how they are weighted.
- Rebalance frequency — monthly and quarterly give materially different results, mostly through cost.
- A regime filter, or none — whether the book moves to cash when breadth deteriorates.
- Costs — charged on every rebalance, not ignored.
Change any one and the result changes. That is why comparing variants matters more than finding a single good number.
The regime filter question
The intuitive addition to any rotation rule is a switch that moves to cash when the market is falling. It usually improves drawdown. Whether it improves the end result depends on how often it is wrong — every false signal costs you a re-entry at a higher price.
The only way to know is to run the same rule with the filter on and off over the same window and look at what it actually cost or saved. An ablation, in other words — and it is worth being suspicious of any rotation strategy presented with its filter already baked in and never tested without.
Doing this without fooling yourself
The Market Pulse scanner shows sector strength and market breadth live, with Mansfield relative strength on every name and stage classification across 2,300+ NSE stocks — the observational half.
For the empirical half, FNOTrader's stock and sector backtesting runs on daily history from 2004 across roughly 2,390 stocks and 17 NSE sector and size indices, and is built survivorship-free: the tradeable universe is rebuilt at every rebalance date from trailing turnover, so companies that were liquid in 2012 and delisted in 2016 are present until they actually died. Signals compute on the close and fill at the next session's open, so there is no same-bar lookahead, and every run reports a regime-filter ablation.
It is reachable by asking, over MCP: "does rotating monthly into the 3 strongest NSE sectors beat holding NIFTY 50 since 2015?" is a question you can put to an assistant connected to your account.
Common questions
What is sector rotation?
The observation that capital concentrates in some sectors and drains from others rather than moving uniformly, and the strategies that try to hold whichever sectors are currently leading. On NSE the observable version is sector relative strength against the broad market.
Why does market breadth lead the index?
An index is a weighted average, so a few large constituents can hold it up while most of the market deteriorates. Breadth measures — advance/decline, share of stocks above the 200-day average, new highs versus lows — show that divergence directly, which the index level cannot.
Why are most sector rotation backtests unreliable?
Three construction faults: survivorship bias from testing on today's constituents, look-ahead from ranking on full-period data then testing from the start, and applying today's sector mapping to past years when membership has changed. All three are properties of how the test was built and cannot be corrected afterwards.
What does point-in-time mean in a rotation backtest?
That at every rebalance date the test uses only information available on that date — including which stocks were liquid and tradeable then. A universe rebuilt at each rebalance from trailing turnover keeps delisted companies present until they actually died, which is what makes the result reproducible.
Does a regime filter improve sector rotation?
It usually improves drawdown; whether it improves the end result depends on how often it signals falsely, since each false signal costs a re-entry at a higher price. The only honest answer comes from running the same rule with the filter on and off over the same window and reading the difference.
How often should a rotation strategy rebalance?
Monthly and quarterly produce materially different results, mostly through accumulated cost. Rather than adopting a convention, run both on the same rule and window and compare after costs — the answer varies by factor and universe.
Related
App: Market Pulse Stocks Scanner · Strategies: Options School (free, no login) · Definitions: trading glossary · Free tools: position sizing & risk calculators · More: all articles