← All articles

Sector rotation, explained simply

· 8 min read
In short

Sectors do not all rise at the same time. What rotation describes, how we measure it, and what our backtest really says about it.

In a market that rises 8%, not every sector rises 8%. Some gain 15, others fall. And the order in which they take the lead is not random: it follows the economic cycle fairly regularly. That is what is called sector rotation.

The idea, without the jargon

When the economy speeds up, investors favour the sectors that benefit from growth: technology, industry, discretionary consumption, raw materials. When it slows, they fall back on what gets consumed whatever happens: healthcare, food, utilities. The first are called cyclical, the second defensive.

The theory is old and well known. What is hard is knowing where we are at a given moment, without telling ourselves the story after the fact.

Strength and momentum: two separate questions

For each sector, we calculate two things from the scores of the stocks in it:

  • Strength: the sector's average score. Where does it stand?
  • Momentum: the change in that score since the day before. Which way is it going?

A sector can be strong and running out of steam, or weak and recovering. The second case is the most interesting, and it is exactly the one an average alone does not show.

The relative rotation graph

The RRG, for Relative Rotation Graph, crosses those two measurements but relative to the market, not in absolute terms. That is the nuance that matters: in a market down 10%, a sector that loses only 3% is leading, even though its performance is negative.

Four quadrants follow:

  • Leading: strong and still improving.
  • Weakening: still strong, but losing momentum.
  • Lagging: weak and sinking.
  • Improving: still weak, but recovering.

In theory, a sector moves through them in that order, clockwise. In practice it skips steps, goes back and hesitates. Hence the question we asked ourselves.

Does it really work? We measured it

A mechanism that looks right on a chart is not proof. So we wrote a backtest that tests three claims separately:

  • Does momentum predict anything? We sort sectors into three groups by momentum and look at the return over the next ten days. Does the top third beat the bottom third, and how often?
  • Do the quadrants really separate? Do Leading and Improving outperform Weakening and Lagging?
  • Is the rotation real? We count the share of transitions that actually go clockwise. If it is close to chance, the cycle is a figure of speech, not a mechanism.

Three measurement precautions, because they change the result. Returns are winsorised at plus or minus 50%: without that, a handful of small stocks that double in ten days decide the verdict on their own. Suspended stocks are excluded by a guard on the gap between dates. And the verdicts rely on the median, not the mean, precisely because the mean gets carried away by extremes.

What the result is worth, and what it is not

The verdict of the latest calculation is shown in the application itself, as a banner on the rotation window, with the median gap and the hit rate. We do not copy it here: a figure in an article ages in silence, while the banner recalculates.

Two limits to keep in mind. The first: we measure rotation on our scores, not on capital flows. It is a defensible proxy, not the thing itself. The second: the reference sector baskets are listed in the United States, so for a European stock the currency gap adds to the sector gap. The page says so rather than burying it.

What it is for in practice

Not for picking a stock. For knowing the environment you are picking in. A good stock in a Lagging sector rows against the current; the same stock in an Improving sector has the wind behind it. It is context, not a signal, and the Sectors page is built for that.

The full method is on Methodology, and the terms used here are in the glossary.

The video tutorial

See all the video tutorials


In the same section
RSI, moving averages, volume: the 5 indicators to know in 5 minutes
Five measurements are enough to read the essentials of a chart: trend, momentum, participation, range and trend strength. None predicts on its own; each one answers a precise question.
How is the ATR stop-loss calculated?
The stop shown is the latest price minus a multiple of the ATR, the average size of the moves over 14 sessions. It is a volatility marker, not an order nor a promise to cap the loss.
What does a candlestick pattern tell you, and what does it not?
A pattern describes the shape of a few candles, never what comes next. The site detects it, shows it, and says what the same shape led to in the past when the measurement allows.

Educational content: these articles explain what the site's measurements calculate, so that you can decide for yourself, knowing what they mean. Investing carries a risk of capital loss. See the legal notice.