The October 2022 print stuck with me. CPI ran 7.7%, the Fed had taken funds from 0 to 3.75% in seven months, the 2s10s spread sat at minus 50 basis points, ISM manufacturing was rolling under 50, and the S&P 500 had given back 25.4% from its January 4 high. Long-duration Treasuries via TLT were down 32% on the year. Energy stocks via XLE were up 59%. Cash, yielding 4.3%, beat both stocks and bonds for the first time in a decade.
That tape was economic cycle asset allocation in action. The four-phase cycle doing what it historically has, just compressed into ten months. The macro regime had flipped from expansion to contraction, and the assets people normally treat as safe sat in opposite corners of the leaderboard from the assets people normally treat as risky. A trader who reads only single-stock charts could not see why every breakout was failing and every defensive rotation was working. The regime had changed underneath the charts.
This is the layer that sits above your screener. Economic cycle asset allocation is the discipline of tilting a book between equities, bonds, commodities, and cash based on where the business cycle is sitting and where it is pointed next. It is not a forecast. It is a posture. And once you understand it, single-stock setups read differently.
The four phases, shaped by money and demand
The cycle has four named phases: expansion, peak, contraction, trough. They describe the path of real GDP relative to its trend, but in trader terms they describe what monetary policy is doing and what end demand is doing.
In expansion, real growth is above trend and accelerating. The Fed is usually accommodative or neutral. Credit spreads are tight. Cyclical sectors lead. This is the dominant phase by duration; in the US between 1945 and 2020, expansions averaged 65 months versus 11 months for recessions, per NBER’s chronology.
At the peak, growth is above trend but decelerating. The Fed is usually mid-cycle to late in a hiking cycle. The yield curve flattens. Inflation runs above target. Inventories build. The peak is not a moment but a window, sometimes 6 to 18 months long, and it is where most macro traders make or lose the bulk of their annual P&L.
Contraction is real growth below trend, often negative. The Fed has begun cutting or is being forced to. Credit spreads widen. Default rates climb. Earnings revisions go negative. The S&P 500 has lost an average 36% peak-to-trough across the 12 NBER recessions since 1945, with a range from minus 14% in 1980 to minus 57% in 2007 to 2009.
Trough is the bottom and the first innings of recovery. Real growth is still negative but second-derivative-positive. The Fed is at or near its terminal cut. Risk assets rip. The cleanest signal in financial history is that the S&P 500 has historically bottomed several months BEFORE the recession officially ends; the March 2009 low arrived a full quarter before the NBER trough date.
Economic cycle asset allocation: what each phase has historically rewarded
The reason this matters for a trader and not just an economist is that returns across asset classes are not stationary across the cycle. They rotate. The Ibbotson and Fama-French datasets, plus more recent work by Fidelity and BlackRock on post-1962 quarterly returns, give a consistent shape:
- Expansion: equities lead, with small-cap and cyclical sectors (industrials, financials, consumer discretionary) at the top. Investment-grade credit returns its coupon plus modest spread compression. Commodities are mixed. Cash is the worst place to be.
- Peak: large-cap quality and energy lead. Commodities and gold tend to outperform on inflation persistence. Bonds suffer because rates are still rising. Defensive equity sectors (staples, healthcare, utilities) begin to outperform cyclicals on a relative basis even while the index is still grinding higher.
- Contraction: long-duration Treasuries lead as rates fall and the curve bull-steepens. Gold often leads inside commodities. Equities fall on average but staples, healthcare and utilities lose less. Cash beats almost everything that is not duration.
- Trough: small-cap equities and high-yield credit lead, often violently in the first six months. Cyclicals reawaken before the data does. Long Treasuries underperform as the curve continues to steepen but for the opposite reason (rate cuts already priced).
The shape is consistent enough that a passive equal-weighted four-asset book that tilts simply by phase has historically outperformed a static 60/40 over the past three cycles on a Sharpe basis. Where the shape breaks is in HOW you decide which phase you are in. I will get to that.
What the shape is NOT: it is not a guarantee that bonds rally in every recession. The 2022 contraction is the cleanest counterexample in 40 years. Long Treasuries fell because the contraction was driven by Fed tightening into inflation, not by demand collapse. Cycle templates are conditional on the kind of contraction, and that is the first place the textbook reading gets traders into trouble.
The four signals I actually watch
You cannot trade off NBER dates. They are announced 6 to 18 months in retrospect. The real-time tools I keep on my desk reduce to four numbers, scanned weekly:
- The 2s10s Treasury spread. Inversion (negative spread) has preceded every US recession since 1969 by 6 to 23 months, with no false positives on the deep-and-sustained reads. I treat sustained inversion past minus 25bp as peak-phase confirmation, not as a contraction signal yet.
- ISM Manufacturing PMI. The 50 level is the demarcation between expansion and contraction in the goods economy. The crossover under 50 has historically led broad equity drawdowns by an average of 4 months, per 1948 to 2022 data. The slope matters more than the level; a three-month rate of change steeper than minus 3 points is the actionable read.
- Initial jobless claims, 4-week moving average. This is the highest-frequency contraction trigger. A 20% rise off the cycle low, sustained for 6 weeks, has historically marked the start of every post-1967 contraction within a quarter.
- BBB-Treasury OAS (option-adjusted spread). Investment-grade credit spreads are the bond market’s vote on default risk. Spreads through 250bp from a cycle low under 150bp have preceded every major equity drawdown since the 1990s by 1 to 4 months.
When two or more of those flip past their thresholds in the same calendar month, I treat the regime as having changed. Not three months later when the data confirms. Then. The tilt happens on the read, not on the confirmation.
A separate book of work uses VIX regime position sizing for the volatility-state dimension of this, and regime factor screening for the equity-factor tilt inside the equity sleeve. Those are sub-layers underneath the macro tilt this article is about, not substitutes for it.
A tilt is not a switch
The most common mistake I see when traders discover cycle work is treating it like an on-off switch: 100% stocks in expansion, 100% bonds in contraction, 100% cash at the peak. That fails for three reasons.
First, the signals are noisy. A 2s10s inversion in July 2022 and a contraction call would have had you out of equities in August 2022, before the October 2022 low and the 26% rally that followed in the four months after. The signal was correct (NBER never called a 2022 recession, but the manufacturing and earnings recession were real); the timing was early, as it almost always is.
Second, the assets do not move on a single date. Defensive sectors begin to outperform cyclicals 6 to 12 weeks before the index tops. Long bonds usually start their rally weeks before the equity peak. Commodity leadership flips even earlier. A tilt that moves 5 to 15 percentage points of exposure across two months matches the slope of the underlying rotation.
Third, the cost of being wrong on a switch is asymmetric. A 100% cash call at a false peak misses every dollar of the final blow-off, which in the late stages of an expansion is often where 15 to 25% of the entire cycle’s index return is concentrated. The 1999 and 2021 endings both put 25%+ on the index after the yield curve had already inverted.
The operator move is to define your peak-phase posture (for me: equity sleeve down 10 to 15 points, defensive sector tilt inside what is left, duration in the bond sleeve creeping out, gold position opened or expanded, cash building), define your contraction posture, define your trough posture, and migrate between them in increments. Each cycle signal that confirms is worth one increment of tilt. Five signals confirmed equals five increments of move. That is the discipline.
Where the textbook breaks down
The cycle framework, like every framework that tries to compress macro into a four-box matrix, has known failure modes. A trader who uses it without knowing them is going to get bagged by the exceptions.
The first is supply-driven contractions. The 1973 to 1974 oil shock, the 2022 inflation shock, and arguably the early-2020 COVID shock all violated the standard “bonds rally in contraction” rule because the contraction had an inflationary cause. The textbook contraction posture (long duration, long gold, short cyclicals) worked partially in 1973 to 1974 (gold yes, bonds no) and failed on bonds in 2022. The lesson is to add an inflation-direction read on top of the phase read. Phase says how to tilt; inflation direction says whether the bond leg of the tilt is even available.
The second is policy-driven distortions. Zero interest rate policy between 2009 and 2015, and again 2020 to 2021, broke the historical correlations between cash, bonds and equities. With cash yielding zero, the cost of defensive positioning was punishing, and many cycle indicators that worked pre-2008 (the yield curve in particular) gave compressed and noisy readings. A cycle template calibrated on 1948 to 2007 data needs a sanity check against the policy regime it is being asked to operate in.
The third is the timing drift between signals and prices. ISM, claims, the curve and credit spreads are all leading indicators. They lead price by different amounts in different cycles, and the lead time is not stable. A 1970 lead of 11 months between curve inversion and recession start has nothing to do with the 2022 lead of 16 months. The honest posture is “the regime is shifting” rather than “the recession starts in November”. The first is tradable. The second is fiction.
The fourth is the difference between a US cycle and a global cycle. Since 2010, the Chinese credit impulse has led global commodity prices by 6 to 9 months, often more cleanly than US monetary policy has. A US-only cycle read missed the 2015 to 2016 industrial recession and the 2019 manufacturing contraction, both of which were China-led. If you trade commodities or any commodity-sensitive equity, the US PMI alone is not enough.
A separate piece on sector rotation by relative strength covers the intra-equity rotation that follows once you have a regime call, and the work of Paul Tudor Jones is the case study most worth reading on combining macro regime reads with disciplined risk increments. The universal portfolios adaptive allocation note covers a math-first variant of adaptive allocation that does not require explicit cycle calls at all.
Cycle reading is a discipline, not a forecast
The economic cycle is not a calendar and it is not a chart pattern. It is the slow grinding of monetary policy and end demand into asset returns over a 4 to 10 year horizon. The traders who use it well treat it as a posture-setting tool, not a market-timing tool. They define what each phase looks like in the book, define the signals that move them between phases, and migrate in increments rather than flips.
The point is not to predict the next recession. The point is to stop running the same equity-heavy book in a peak that you ran in an expansion, and to stop running the same defensive book in a trough that you ran in a contraction. The cycle does not tell you what to buy. It tells you what posture is rewarded for the regime you are in. Then the chart work does its job inside that posture.
Learn the pattern. Ride the trend. Keep the gains.
Educational content only. Not investment advice. Trading involves risk. You are responsible for your decisions.