Sector Rotation and the Business Cycle: Which Sectors Lead in Each Phase, and How to Tilt Without Guessing

On September 30, 2026, the Bureau of Economic Analysis published its third estimate of second-quarter GDP: real growth of 2.2% at an annual rate, with the first quarter revised to 2.5%. That is solid, useful information about April, May and June, delivered three months after June ended. The first estimate of the third quarter is not due until October 29. Every investor who wants to rotate between stock sectors as the economy turns runs into the same problem: the economy reports on itself late, revises itself often, and the official verdict on whether a recession has started can take a year. Sector rotation is still worth understanding, because the sectors that lead and lag across the cycle tell you where your portfolio is exposed. It is just not worth betting the portfolio on. If you are deciding how much room tactical moves should get at all, start with strategic vs tactical asset allocation; this guide is about the sector piece.

What is sector rotation, and how does it tie to the business cycle?

Sector rotation is shifting stock exposure between market sectors as the economy moves from recovery to expansion to slowdown to recession, overweighting the sectors whose profits are most sensitive to the phase ahead and underweighting the rest. It rests on a simple observation: a bank, a steelmaker, a chipmaker and a toothpaste company do not earn their money on the same schedule.

The sectors themselves are standardized. The Global Industry Classification Standard (GICS), developed by MSCI and S&P Dow Jones Indices, sorts listed companies into 11 sectors: Energy, Materials, Industrials, Consumer Discretionary, Consumer Staples, Health Care, Financials, Information Technology, Communication Services, Utilities and Real Estate. Most sector ETFs and index funds follow those definitions, which is what makes rotation practical to implement.

Investors group those 11 into two families. Cyclical sectors — financials, industrials, consumer discretionary, materials, energy and, in most frameworks, technology — have revenue and margins that swing with the economy. Defensive sectors — consumer staples, utilities and health care — sell things people keep buying in a downturn. The rotation thesis is that you want more cyclical exposure when growth is about to accelerate and more defensive exposure when it is about to fall.

The reason this links to the cycle rather than to last quarter’s GDP is that the stock market looks ahead. The S&P 500 is itself one of the ten components of The Conference Board’s Leading Economic Index, precisely because prices tend to turn before the economy does. Sector leadership turns earlier still. By the time a phase is obvious in the data, the sectors that benefit from it have usually moved. Rotation is therefore a forecasting exercise, and it should be sized like one.

What are the four phases of the business cycle?

Investors describe four phases — early, mid, late and recession — but the official referee of the US cycle only dates two points: peaks and troughs. The National Bureau of Economic Research (NBER) defines a recession as “a significant decline in economic activity that is spread across the economy and lasts more than a few months,” and judges it on depth, diffusion and duration. Its committee gives the most weight to real personal income less transfers and nonfarm payroll employment, alongside real consumer spending, manufacturing and trade sales, household employment and industrial production.

The four investing phases are a convention layered on top of those dates:

• Early cycle (recovery). Growth rebounds from the trough, often sharply. Inflation is low, the Federal Reserve has cut rates, credit starts to loosen and the yield curve is steep. Corporate earnings recover from a depressed base, so the most economically sensitive businesses see the biggest percentage gains.
• Mid cycle (expansion). The longest phase. Growth is positive but moderating, hiring is steady and the Fed moves toward neutral. Earnings growth is broad and less dramatic, and business investment in equipment and software picks up.
• Late cycle (slowdown). The economy is running near capacity, labor is tight, inflation pressure builds and the Fed has tightened. The yield curve flattens or inverts. Margins come under pressure and earnings growth narrows.
• Recession (contraction). Output, income and employment fall. Inflation cools, the Fed cuts, and the curve re-steepens as short rates fall faster than long ones. Earnings decline, most in the most cyclical businesses.

Two caveats matter more than the labels. First, phases are not equal in length or guaranteed to arrive in order: the economy can stall in a mid-cycle slowdown and re-accelerate without a recession, and it can skip almost straight from expansion to contraction after a shock. Second, the phase you care about is the next one, because that is what prices are discounting.

Which sectors tend to lead in each phase of the cycle?

Conventionally, financials, consumer discretionary and industrials lead early; technology, communication services and industrials lead mid-cycle; energy, materials and health care lead late; and consumer staples, utilities and health care hold up best in recession. The logic in each case is about whose earnings are about to change most, not whose earnings are highest today.

Read Chart 1 down a column to see the macro fingerprint of a phase, or across the leaders row to see how leadership moves from rate-sensitive cyclicals to growth to inflation beneficiaries to defensives. The table puts the reasoning next to each tilt.

Notice what the map says about defensive sectors in a recession. Their job is to fall less, not necessarily to rise. A staples overweight that loses less than the index during a contraction has done its job even if it loses money. Notice too how much the map depends on the reason for the cycle. A recession caused by an oil shock, a banking crisis or a pandemic will punish different sectors than one caused by the Fed squeezing out inflation. The framework tells you which questions to ask; it does not answer them.

How long do US expansions and recessions usually last?

Since the Second World War, expansions have averaged 64.2 months and recessions 10.3 months, using NBER reference dates for 1945 to 2020. That asymmetry is the single most important fact for anyone tempted by a defensive rotation: the economy spends roughly six months expanding for every month it contracts.

The spread around those averages is wide. The expansion that ended in February 2020 ran 128 months, the longest on the NBER’s record, and was followed by the shortest recession, just 2 months to the April 2020 trough. The 1980 recovery lasted only 12 months before the July 1981 peak. The December 2007 to June 2009 contraction ran 18 months, the longest of the post-war era. Twelve cycles is a small sample, and no two have the same cause.

Three practical conclusions follow for sector positioning:

1. Defensive rotation is expensive insurance. If you overweight staples and utilities on a recession call that is a year early, you can give up a meaningful share of a long expansion to protect against a contraction that, on average, lasts under a year.
2. Most of the action is inside the expansion. With expansions averaging more than five years, the early-to-mid and mid-to-late transitions are the rotations you will face most often. Treat them as more important than calling the recession itself.
3. Recoveries come fast. Because contractions are short, the early-cycle window is short too. Investors who wait for the all-clear tend to miss the part of the cycle the framework says pays best.

None of this means recessions are harmless; drawdowns in a contraction are what force people to sell at the worst time. That risk is better handled by how the whole portfolio is built than by a well-timed sector switch. Our recession-proof portfolio guide covers the structural side.

Why can’t you just wait for an official recession call?

Because the official call is a historical verdict, not a signal. The NBER says its approach to dating turning points is retrospective and that its committee waits until enough data are in to avoid major revisions. Since 1990, it has announced business-cycle peaks 4 to 12 months after they happened and troughs 15 to 21 months after.

Take the 2007–2009 recession. The peak was December 2007; the NBER announced it on December 1, 2008, after the worst of the financial crisis had already hit markets. The trough was June 2009; the announcement came on September 20, 2010. An investor who waited for the first announcement to rotate defensively would have done so near the end of the contraction, and one who waited for the second to rotate back into cyclicals would have sat out more than a year of recovery.

Look at the peak announcements alone. In 1990–91 the peak was confirmed in April 1991, a month after the recession had already ended. In 2001 the announcement came in November, the very month of the trough. In 2007–09 it came 12 months into an 18-month recession. Only in 2020, with a collapse no one could miss, was the call made quickly. GDP has the same problem on a smaller scale: the first estimate of a quarter arrives about a month after it ends and is revised twice more, as the September 30 release shows. The data you need to rotate on has to come from somewhere else.

Which indicators tell you where we are in the cycle?

Use leading indicators to anticipate the next phase and coincident data to confirm the current one: The Conference Board Leading Economic Index, the 10-year minus 3-month Treasury spread, weekly jobless claims, monthly payrolls and the direction of Federal Reserve policy. No single series is reliable alone, which is why a rotation rule should require several to agree.

The Leading Economic Index (LEI). The Conference Board combines ten series that tend to turn before the economy: average weekly hours in manufacturing, initial claims for unemployment insurance, manufacturers’ new orders for consumer goods and materials, the ISM new orders index, new orders for nondefense capital goods excluding aircraft, building permits, the S&P 500, its Leading Credit Index, the 10-year Treasury less federal funds spread, and consumer expectations for business conditions. Its published “3Ds” rule signals an impending recession when the six-month diffusion index is at or below 50 and the LEI’s six-month annualized growth rate falls below −4.3%, both at once. Requiring both depth and breadth is what keeps the rule from firing on a single noisy month.

The yield curve. In their 1996 Federal Reserve Bank of New York study, Arturo Estrella and Frederic Mishkin found that the spread between the 10-year Treasury note and the 3-month Treasury bill outperformed other financial and macroeconomic indicators in predicting recessions two to six quarters ahead. An inverted curve says markets expect short rates to fall, which usually happens because the economy weakens. It is a slow signal with a variable lead; the curve can stay inverted for many months before a downturn, and the re-steepening that follows an inversion is often closer to the recession itself. Our yield curve explainer covers the 10-year minus 2-year versus 10-year minus 3-month debate and the lag in detail, and the economic indicators guide for executives shows how the curve fits alongside the other leading series.

Labor data. Initial jobless claims are weekly and among the fastest real-economy reads available; payrolls are monthly and are one of the two series the NBER weights most. A sustained rise in claims is a late-cycle warning; payroll declines across many industries are the recession itself.

The Fed. The Federal Open Market Committee holds eight regularly scheduled meetings a year. The direction of policy — hiking, on hold, cutting — is often the clearest marker between late cycle and early cycle, because the textbook early phase begins with easy money.

Market signals add a useful cross-check. If cyclical sectors are leading but only a handful of large stocks are carrying the index, the move is fragile; market breadth indicators show whether participation confirms what the macro data say.

Does sector rotation actually beat buy-and-hold?

Not reliably once hindsight is removed. The framework describes average tendencies that are clear in the rear-view mirror, but the gains shrink sharply when you have to identify the phase in real time and pay to trade. Academic tests such as Stangl, Jacobsen and Visaltanachoti’s “Sector Rotation over Business Cycles” are often cited for exactly that finding: even assuming you know the official phase dates in advance, the outperformance from textbook rotation is modest, and it fades once you allow for not knowing the dates and for transaction costs.

Four things erode the edge in practice:

• Timing error. Charts 2 and 3 show the problem from both sides: short recessions leave little room for error, and the official dates arrive too late to trade. A rotation that is right about the phase but six months early or late can lose money.
• Leadership that breaks the template. Each cycle has its own cause. In 2020 the contraction lasted two months and the leaders were the companies that benefited from a locked-down economy, not the sectors the textbook would have picked. Structural trends, such as the long rise of software and semiconductors, can overwhelm cyclical patterns for years.
• Costs and taxes. Each rotation in a taxable account can realize gains. Gains on positions held one year or less are short-term and taxed at ordinary-income rates under IRS rules, which can turn a small pre-tax edge into an after-tax loss.
• Crowding. The phase map is public. When everyone expects the same rotation, it gets priced early, which is the same reason contrarian timing tends to beat consensus timing.

That does not make the framework useless. It makes it a risk map: it tells you which of your holdings are most exposed to the next phase and lets you lean modestly toward the ones that are not. That is a very different job from trying to own only next year’s winning sector. Rules-based approaches that move gradually with the data, discussed in our guide to dynamic asset allocation, are a better fit for that job than big discretionary switches.

How should you size a sector rotation tilt?

Size tilts small, rules-based and relative to your strategic weights: a common Proflex approach is to let any one sector move no more than about 5 percentage points from its strategic weight, and only when two independent signals agree. The goal is to improve the odds a little without letting a wrong cycle call dominate the portfolio’s result.

The decision tree in Chart 4 is deliberately conservative. It starts with the LEI’s 3Ds rule because it is a published, mechanical threshold that already requires breadth and depth. It then asks the yield curve, a market-based signal with a different source of information, to confirm. Only when both point the same way does the tilt reach its full size. When they disagree, the default is the mid-cycle answer: hold your strategic weights and rebalance at the bands.

Three implementation choices make the tilt easier to live with:

• Use a core and a satellite. Keep most of the equity allocation in a broad, low-cost core at strategic weights and run tilts through a smaller satellite sleeve of sector funds. If the call is wrong, the damage is limited to the satellite.
• Move in steps. Shift a third of the planned tilt at each confirmation (signal fires, signal persists the next month, second signal agrees). Staging reduces the cost of false signals, which the LEI and the curve both produce.
• Write the exit before the entry. Decide in advance what would reverse the tilt: the 3Ds rule switching off, the curve normalizing, claims falling back. Rotations fail most often because investors keep the defensive tilt long after the recovery has started.

Volatility matters too. Sector funds are narrower than the index and some are far more volatile, so a 5-point tilt into energy carries more risk than a 5-point tilt into staples. One simple fix is to size each tilt by risk rather than by dollars: a smaller position in a volatile sector can carry the same risk as a larger one in a steady sector.

How do you rotate a concentrated tech position without a big tax bill?

Start by recognizing that a concentrated position is already the biggest sector bet in the portfolio, then rotate around it with new cash, vesting shares, options and tax-advantaged accounts before you sell low-basis stock. For many Proflex readers — engineers and executives with employer stock or RSUs — the portfolio’s cycle exposure is set mostly by that one holding.

Consider a hypothetical household with a $3 million portfolio: $1.2 million in one large-cap semiconductor company received through RSUs, at a cost basis of $200,000, and $1.8 million in a broad index fund. The index fund itself carries a substantial technology weight, so the household’s effective Information Technology exposure is well above half the portfolio before any rotation decision. A defensive rotation that adds 5 points of staples while leaving the single stock untouched barely changes the risk.

In this example, the household could stop the position growing by selling every vest, route all new savings to underweight sectors and run any cycle tilts inside the retirement accounts. If the decision tree points to a late-cycle or defensive stance, a collar on the concentrated stock sets a floor under the largest cyclical exposure for a defined window without triggering a sale. Selling low-basis shares then becomes a deliberate, staged decision rather than a reaction to a recession headline. Our guide to taxes on sold stock walks through how the gain on those shares is calculated.

What does a disciplined sector rotation routine look like?

A workable routine is a quarterly review with written rules, a short list of signals and hard caps on tilt size. It is boring on purpose: the value comes from consistency, not from catching every turn.

1. Write down your strategic sector weights. Usually the weights of a broad index, adjusted for any concentrated position you hold. Every tilt is measured from here.
2. Check the leading signals. LEI and its 3Ds status, the 10-year minus 3-month spread, the four-week average of jobless claims and the latest FOMC decision. Note which phase each one points to.
3. Count agreement. Tilt only when at least two independent signals point the same way. One signal on its own means no change.
4. Size from the decision tree. Apply the tilt in steps, capped at about 5 points per sector, inside the satellite sleeve.
5. Cross-check with the market. Look at which sectors are already leading and whether breadth confirms it. If the move is already crowded, scale the tilt down.
6. Put trades in the right account. Run rotations in tax-advantaged accounts first; in taxable accounts, prefer new money and vest proceeds over selling appreciated shares.
7. Record the exit trigger. Write what would reverse the tilt, and check it next quarter before adding anything new.

Run this way, sector rotation becomes a modest, repeatable edge on top of a sound plan: you know which phase each part of the portfolio is betting on, you lean a little toward the next phase when the evidence lines up, and you never need the NBER to tell you what already happened.

Frequently asked questions

Key takeaways

1. Sector rotation tilts a portfolio toward the sectors that conventionally lead in each business-cycle phase; it is a framework, not a law.
2. The US cycle is lopsided: NBER data show expansions averaging 64.2 months and recessions 10.3 months from 1945 to 2020, so most of your time is spent in early, mid and late expansion.
3. Official turning points arrive late: the NBER confirmed peaks 4 to 12 months after the fact and troughs 15 to 21 months after, from 1990 to 2021.
4. Read the cycle from leading data instead: the LEI and its 3Ds rule, the 10-year minus 3-month Treasury spread, jobless claims and the direction of Fed policy.
5. Require two independent signals to agree before tilting, and cap any sector tilt at about 5 percentage points of the portfolio.
6. For concentrated tech holders, the biggest sector bet is the one you already own; rotate with new money, vests and options before you sell low-basis shares.
7. Rotation costs taxes and trading friction; make taxable moves rarely, and use tax-advantaged accounts for the active sleeve where you can.

Sources and method

1. NBER, US Business Cycle Expansions and Contractions — peak and trough dates and the 1945–2020 average durations (64.2 months of expansion, 10.3 months of contraction) used in Chart 2 and the stat strip
2. NBER, Business Cycle Dating Committee Announcements — announcement dates used for the lag figures in Chart 3 and the turning-point table
3. NBER, Business Cycle Dating — the definition of a recession and the indicators the committee weighs
4. Estrella and Mishkin, The Yield Curve as a Predictor of U.S. Recessions (Federal Reserve Bank of New York, 1996) — why the 10-year minus 3-month Treasury spread is the curve measure to watch, with a horizon of two to six quarters
5. U.S. Bureau of Economic Analysis, Gross Domestic Product — the Q2 2026 third estimate (2.2% annualized) and GDP release timing

Durations and announcement lags are taken directly from the NBER pages listed, as accessed on October 1, 2026; lags are counted in calendar months from the turning-point month to the announcement month. The phase-to-sector map (Chart 1 and the first table) is the conventional business-cycle framework used across the industry, summarized by Proflex; it is not a backtest and no sector return figures are claimed. The ten LEI components and the 3Ds rule (six-month diffusion index at or below 50 and six-month annualized LEI growth below −4.3%) are taken from The Conference Board’s US Leading Indicators page (conference-board.org/topics/us-leading-indicators) and its Description of Components page, both accessed October 1, 2026; they are cited in text rather than linked because the Conference Board server does not answer automated link checks. The decision tree, tilt sizes and the worked example are illustrative Proflex frameworks with hypothetical numbers. Tax points refer to general federal rules; this is education, not tax or legal advice; confirm with your CPA or counsel.