Sector rotation, the business cycle, and when correlations break

Money moves between sectors before it leaves the market altogether. Watching which sectors lead is a way of reading the market's opinion about the economy, and it is also the last piece of the regime sentence. The lesson ends with the most important intermarket skill of all: recognising when a relationship you have been relying on has stopped working.

The classic rotation model

The textbook sequence runs like this, with the equity market leading the economy by six to twelve months:

        early        mid          late        recession      early
   ───────┬────────────┬────────────┬────────────┬────────────┬───
  Fin  ▓▓▓▓▓                                              ▓▓▓▓▓
  Disc ▓▓▓▓▓▓▓                                              ▓▓▓
  Tech       ▓▓▓▓▓▓▓▓▓▓▓
  Indu    ▓▓▓▓▓▓▓▓▓▓
  Ener              ▓▓▓▓▓▓▓▓
  Matl              ▓▓▓▓▓▓▓
  Stpl                    ▓▓▓▓▓▓▓▓▓▓
  Util                        ▓▓▓▓▓▓▓▓▓▓
  Hlth                        ▓▓▓▓▓▓▓▓▓

This model is a useful map and an unreliable clock. Cycles are irregular, sector definitions change (technology is a far larger share of the index than when the model was drawn, and "technology" now includes companies that behave like staples), and policy interventions distort the sequence. Use it to interpret what you see, not to predict what comes next.

Ratios that summarise it

Instead of tracking eleven sectors, a few ratios capture the offence-versus-defence question directly:

Two or three of these on a weekly chart give you a compact rotation read. When most are rising with the index, the rally has the cyclical, risk-seeking character of an expansion. When the index rises while most fall, the rally is defensive or narrow, which is how late-cycle advances tend to look.

Key idea: Sector leadership reveals what the market thinks about the economy. Offence-versus-defence ratios summarise it; when they diverge from the index, the character of the rally has changed even if the price has not.

Rotation within a bull market

Not every rotation is a cycle turn. In a healthy uptrend, leadership passes from one group to another every few weeks as the crowd chases and abandons themes. What distinguishes healthy rotation from distribution:

• Healthy: money leaves one sector and shows up in another; breadth stays broad; the equal-weight ratio holds up.
• Distribution: money leaves one sector and shows up in defensives or cash; breadth narrows; new lows expand in the abandoned group and do not contract elsewhere.

The Module 3 tools tell you which is happening. Sector ratios tell you where.

When correlations break

Every relationship in this module has failed at some point. The stock-bond correlation flipped in 2022. The dollar and commodities rose together during supply shocks. Crypto decoupled from tech for months at a time. The yield curve inverted and stocks rallied for a year. The rotation model mis-ordered energy and technology for most of a decade.

This is not a reason to abandon intermarket analysis. It is a reason to hold every relationship as a hypothesis with a measured confidence, and to notice when the data stops supporting it.

How to notice. For each relationship you use, keep a rolling correlation (60 days for daily data, 26 weeks for weekly). Set two thresholds: one for "relationship is holding" and one for "relationship has broken". For example, stock-bond correlation: above +0.3 is a growth regime, below -0.3 is an inflation regime, between them is "not reliable right now, do not use".

What a break usually means. A correlation breaks when the dominant driver of the market changes. Growth to inflation. Domestic to global. Fundamental to liquidity. The break is itself information: it says the old regime sentence is out of date and a new driver has arrived. The right response is to ask what the new driver is, not to force the old relationships back onto the data.

What to do while it is broken.

1. Drop the broken relationship from the regime sentence. Say "stock-bond correlation unstable" instead of guessing.
2. Lean harder on the observed internals from Modules 2 and 3. Breadth and volume are measured, not inferred; they do not "break" the way a correlation does.
3. Reduce size in strategies that assumed the relationship. A hedge that no longer hedges is an unhedged position.
4. Wait for the rolling correlation to re-establish itself for several weeks before trusting it again.

The closing thought for intermarket work

The stock market's internals are a census: every stock is counted. Intermarket relationships are inferences: this market usually moves with that one. A census can be misread, but it does not stop being true. An inference can simply stop applying. That is why this course put internals first and intermarket last, and why the dashboard in the next module weights them accordingly.

Try it: Pick the three intermarket relationships you find most convincing. For each, compute the rolling 60-day correlation over the last five years and count how many months it spent in the "broken" zone you define. Write the count next to the relationship in your journal. That number is how much to trust it.

Recap

• The classic rotation model maps sectors to cycle phases; it is a useful interpretation tool and an unreliable predictor.
• Offence-versus-defence ratios (discretionary/staples, semis/index, small/large, high-beta/low-vol) summarise rotation in a few lines.
• Healthy rotation moves money between sectors with breadth intact; distribution moves it to defensives with breadth narrowing.
• Every intermarket relationship has broken at some point; track rolling correlations and drop a relationship from the regime sentence when it is outside its holding zone.
• Observed internals do not break the way inferred correlations do, which is why they come first.