Market regime today

Where each market sits right now

Regions routinely sit in different phases at the same time, so a single United States reading stretched over the rest of the world is misleading. CORVIX classifies all seven markets it covers separately, on the same rubric, and the table below is the current call for each. It is the same classification the app uses, refreshed on the same schedule.

The phase is the call. The five-category scoring behind each call, the data source driving each category, and the confidence attached to it are in the app.

What a market regime is, and why it is worth tracking

Most portfolio outcomes are decided less by which individual holdings you picked than by the backdrop you held them through. A market regime is that backdrop: the combination of where the economy sits in its cycle and what growth and inflation are doing to each other. The same balanced portfolio can compound quietly for years in one regime and lose a third of its value in another, without a single holding changing.

CORVIX describes the regime two ways at once, because the two answer different questions. The business cycle phase answers where in the expansion or contraction the economy currently sits. The macro quad answers what growth and inflation are doing together right now, which is usually the faster-moving of the two. Both readings are shown above, alongside a correction risk estimate that measures how much stress is visible in markets today.

The four business cycle phases

Each region is scored on two axes: the level of economic activity, and the momentum of that activity. Those axes are built from five categories, namely growth, credit, corporate profits, policy stance, and inventories. Where a region lands on the two axes places it in one of four phases.

The four macro quads

The quad is a faster-moving read than the cycle phase. It asks only what growth and inflation are each doing, accelerating or decelerating, and crosses the two answers. It matters because the correlation between stocks and bonds is not fixed: it depends heavily on which of these four boxes the economy is in.

What correction risk is measuring

The correction risk figure is a probability estimate built from three inputs, each of which has historically moved ahead of or alongside equity drawdowns. A higher reading means more of the three are flashing at once.

The three are combined through a logistic function, which keeps the output inside a sensible probability range and stops any single input from dominating. One important consequence of the design is worth stating plainly: each input is scored against fixed thresholds rather than continuously, so the reading moves in visible steps rather than drifting smoothly. A spread crossing one of those thresholds shifts the figure more than a large move that stays inside a band. That is a deliberate trade for interpretability, and it means small changes in the number are not meaningful.

The suggested cash weighting shown alongside it is derived directly from this figure by a fixed published mapping, so it is not a separate opinion. It is the same reading expressed as a position size.

Two honest caveats. It carries no timing information, so a high reading can persist for months without a correction arriving, and a correction can begin from a low reading. And it is an estimate rather than a forecast: it describes how much stress is visible now, not what will happen next. It is most useful as a check on your own positioning rather than as a trigger.

How often this reading changes

The numbers above are refreshed on a schedule through the day from the same engine that powers the CORVIX app, and the exact timestamp of the current reading is shown directly beneath them. The phase itself moves far more slowly than the timestamp suggests: because the underlying economic data is released weekly and monthly rather than continuously, a phase change usually takes weeks or months to develop. The quad and the correction risk reading move faster.

CORVIX classifies all seven regions it covers separately rather than stretching a single United States reading over the rest of the world, since regions routinely sit in different phases at the same time. The reading above is the global summary. The full per-region board, the theme-by-theme breakdown, and the point-in-time backtester are inside the app.

The five categories behind each phase call

A phase label is only as good as what sits underneath it. Each region is scored on five categories, and every category is calculated from published data rather than judgement. Where a category has no clean data source, CORVIX uses a stated proxy and says so on the page rather than quietly filling the gap. The categories are not equally weighted: growth carries the most, inventories the least, and the weights are fixed rather than tuned after the fact.

How the categories become a phase

The five categories feed two axes. The first is the level of activity, which asks whether the economy is running above or below trend right now, and is built from the growth category alone. The second is the momentum of activity, which asks whether conditions are improving or deteriorating, and is built from leading indicators together with the credit, profits, policy and inventories categories.

Crossing the two axes produces the four phases. Below trend and improving is Early. Above trend and improving is Mid. Above trend and deteriorating is Late. Below trend and deteriorating is Recession. That is the whole classification: there is no discretionary override and no analyst adjusting the answer afterwards.

Two details matter for reading the output honestly. First, a small dead band sits around each boundary, so a region hovering on the line between two phases does not flip back and forth on noise. A decisive move across a boundary takes effect immediately; a marginal one has to hold. Second, every category is weighted, but the weights are renormalised over whatever data is actually available. If a region is missing its credit series, that category drops out and the remaining four are scaled back up to fill the gap, rather than the missing input being silently treated as zero. The proportion of the possible weight that was genuinely present becomes a confidence figure, shown alongside the phase in the app.

When a region has no growth data at all, CORVIX holds the previous phase and marks the reading low confidence, rather than guessing a phase from momentum alone. A stale answer that is labelled stale is more useful than a fresh answer that is wrong.

Why regions sit in different phases at the same time

Divergence is the normal state, not the exception, and it is the main argument against reading a single United States number as though it described the world. Three mechanisms drive it.

Central bank cycles are not synchronised. One region can be cutting into a slowdown while another is still raising into an inflation problem. Because policy stance is one of the five categories, the same nominal rate produces a completely different score depending on where each region's inflation sits.

Credit cycles are largely domestic. Bank lending standards, household leverage and property cycles are set by local conditions and local regulators. A credit contraction in one economy can run alongside a credit expansion in another for years, and the credit category picks that up directly.

Commodity importers and exporters experience the same price move in opposite directions. An oil move that squeezes margins and household budgets in an importing economy lifts national income in an exporting one. The growth and profits categories then move opposite ways for the same global event.

The practical consequence is that a portfolio holding international exposure is holding several cycles at once. Whether that is diversification or concentration depends on whether those cycles are actually in different places, which is a question the per-region board answers directly and a single global reading cannot.

How this compares to the frameworks you already know

Business cycle analysis is not new, and several well-established frameworks already describe roughly the same thing. It is worth being explicit about how this reading relates to them, because the differences are the reason it exists rather than marketing points.

The NBER recession dating committee is the official arbiter of when a United States recession began and ended. It is also deliberately, unavoidably retrospective: the committee waits for data to settle and revisions to land before it announces anything, and its announcements have historically arrived many months after the turning point they describe. That is the correct design for an official historical record and the wrong tool for a decision you are making now. A Recession reading on this page is emphatically not an NBER call, and the two can disagree for a long time in both directions.

The Conference Board Leading Economic Index is a composite of ten forward-looking series and is a genuinely useful indicator. Two things limit it here. It publishes monthly rather than continuously, and the headline versions cover a small number of large developed economies, so it cannot answer the per-region question at all. Where a leading index is available for a region, CORVIX uses it as one input into the momentum axis rather than as the answer.

Purchasing managers' indices are the single most widely quoted cycle indicator, and for good reason: they are timely, survey-based and available across most economies. They are also one survey. Treating a PMI print as the cycle is how a manufacturing inventory wobble gets read as a recession. Here it is one component of one of five categories.

House business cycle frameworks published by large asset managers use much the same four-phase shape, and where this page uses Early, Mid, Late and Recession it is deliberately speaking the same language rather than inventing a private vocabulary. The difference is what sits behind the label. A house view is a considered judgement produced by a team, updated on their schedule, with the inputs and weightings generally not disclosed. This reading is a computation: the inputs are named, the weights are published, no analyst adjusts the output, and the same rules can be run over history to see what they would have said at the time.

That last point is the one worth dwelling on. A framework you cannot backtest is a framework you have to take on trust. Because every input here is point-in-time and every weight is fixed rather than fitted after the fact, the same classification can be replayed across past periods and judged on what it actually said, not on what it would say with hindsight. That does not make it right. It makes it checkable, which is a different and more useful property.

What each phase has historically favoured

The reason a phase label is worth computing at all is that different parts of the market have tended to behave differently in each one. These relationships are the oldest and most studied part of cycle analysis, and CORVIX encodes them as a published table rather than leaving them implicit. The table below is the actual set of tilts the app applies, not an illustration of the idea.

Each cell is a fixed score between -1 and +1 in the published table; the wording above groups the strong positive and strong negative entries.

The economic reasoning behind each row is straightforward, and it is worth understanding rather than memorising.

Early is the phase where credit reopens and demand that was postponed gets spent. Financials benefit directly because lending volumes recover from a low base and loan losses stop growing. Consumer Discretionary and Real Estate benefit because purchases people deferred during the downturn are made, often financed. Industrials benefit as capital spending restarts. What is out of favour is what was defensive: Utilities and Consumer Staples kept their earnings through the bad period and are correspondingly less interesting when everything else is recovering faster.

Mid is the longest and least distinctive phase, which is why its tilts are the mildest of the four. Growth is established rather than accelerating, and the sector picture is correspondingly flatter. Technology and Communication Services tend to do relatively well when capital spending is healthy and there is no immediate pressure on margins.

Late is where the picture inverts. Capacity is tight, input costs and wages are rising, and central banks are usually tightening into that. Energy and Materials sit on the profitable side of rising input costs rather than the paying side. Staples and Healthcare become interesting again because demand for them does not depend on the cycle. Financials come under pressure from a flattening or inverted curve, and rate-sensitive Real Estate and Consumer Discretionary feel higher financing costs directly.

Recession concentrates into the sectors whose revenue barely moves with the economy. People do not stop buying food, taking medicine or heating their homes. Everything that depends on discretionary spending, capital investment or credit growth is on the other side of that.

Three caveats matter more than the table. First, these are tendencies observed across many cycles, not rules, and any individual cycle can contradict them completely — a sector can be cheap enough or expensive enough to overwhelm its phase relationship. Second, CORVIX does not act on the tilt alone: it is one component combined with live valuation, momentum, trend and crowding signals, so a sector the phase favours can still rank poorly today. Third, and most importantly, the tilt is applied to relative positioning within a book, not to a decision about whether to be invested at all.

Reading the page when the signals disagree

The most common source of confusion is that the figures on this page can point in different directions at once. That is not a malfunction. They measure different things over different horizons, and the disagreements are frequently the most informative thing available.

The phase says Late but the quad looks benign. The phase describes where activity sits in the cycle; the quad describes the current combination of growth and inflation. An economy can be late in its cycle while inflation is falling and growth is holding up, which is exactly what a soft landing looks like while it is happening. The two are not in conflict; one is a position and the other is a condition.

Correction risk is elevated but the phase is Mid. Correction risk is built from credit spreads, the change in policy rates and market breadth, and all three can deteriorate well before the economic data does. Markets price expectations; the phase reads released statistics. This particular disagreement — market stress rising while the economy still looks fine — is the ordinary shape of the early part of a turn, and it is also the ordinary shape of a false alarm. It does not distinguish between them.

Regions are in different phases. Treat this as information rather than noise. If a portfolio is spread across regions that are genuinely in different parts of their cycles, that is diversification doing what it is supposed to do. If the regions have converged into the same phase, apparent geographic diversification is not buying much.

The confidence figure is low. This means a meaningful share of the inputs were unavailable and the remaining categories were scaled up to compensate. The phase shown is the best reading from what was present, but it is resting on fewer legs than usual, and it deserves proportionally less weight. A low-confidence phase change in particular is worth waiting out rather than acting on.

How a regime model fails, and how you would know

Every framework of this kind has characteristic failure modes. Naming them is more useful than claiming they do not exist.

Data gets revised. GDP and employment figures are routinely restated, sometimes substantially, months after first publication. A phase computed from first prints can therefore be classified differently once the revised data arrives. This is unavoidable for anything that wants to be timely, and it is a genuine argument against treating a fresh reading as settled. It is also the reason the backtester uses point-in-time data: replaying history with today's revised figures would make any model look far better than it was.

Transitions are messy, and the middle of one is where the model is weakest. The classification is cleanest when a region sits well inside a quadrant and least reliable when it sits near a boundary. The dead band around each boundary reduces flapping but does not create certainty. A reading that has just changed phase is the least trustworthy reading the model produces.

The proxies can break. Corporate profits are stood in for by equity momentum, and inventories by a construction from the PMI. When equities are badly wrong about future earnings — which happens, in both directions, around inflection points — the profits category is wrong with them, and it is wrong in a correlated way with the market rather than independently. That is the least comfortable dependency in the whole framework and it is disclosed for that reason.

Fixed weights are a deliberate constraint, and constraints cost something. The weights here are set from reasoning and left alone, rather than tuned to whatever would have performed best historically. That guarantees the model is not fitted to the past, and it equally guarantees it is not optimal for any particular period. It is the right trade for a framework meant to be honest rather than impressive, but it means a well-tuned model will beat it over any given stretch of history.

What would count as the framework being wrong. Not a single incorrect call — any classifier will make those. The things that would genuinely falsify it are structural: phases that flip repeatedly without the underlying data moving decisively, the per-region readings converging so consistently that separate classification adds nothing over a single global number, or the sector tilts showing no relationship to outcomes when replayed across history. Those are testable, and the backtester exists so that they can be tested rather than argued about.

How this page is actually used

The reading is a starting point for a question, not an output to act on. In practice it gets used in three ways, and being explicit about them is more useful than leaving it abstract.

As a consistency check on positioning you already hold. The most common use is the least dramatic: look at the phase, look at what you own, and ask whether the two still match the reason you bought it. A portfolio assembled during an Early reading and never revisited is now sitting in whatever phase has arrived since, which may or may not still suit it. The value here is that the question gets asked on a schedule rather than after something has already gone wrong.

As a check on whether geographic diversification is real. Holdings spread across several regions are only diversified across cycles if those regions are actually in different phases. The per-region board answers that directly, and the answer changes over time. Regions converging into the same phase is one of the few things on this page that has a fairly unambiguous reading.

As context for news, rather than a substitute for it. A headline about rate cuts means something different in an Early reading than in a Late one. Knowing where the cycle sits does not tell you what will happen, but it does tell you which of several plausible interpretations of the same event is more consistent with the rest of the data.

What it is deliberately not built for is short-term timing. Nothing here updates fast enough or carries enough forward information to support a decision measured in days, and treating a slow-moving structural reading as a trading signal is the most reliable way to be disappointed by it.

What this page does not tell you

It is not a forecast. Every figure describes conditions that are already observable in released data, and none of it is a prediction of what those conditions will do next. The phase is a description of where the economy has got to, not a statement about where it is going.

It carries no timing. A Late reading can persist for a long time, and the historical record contains long stretches of strong returns inside late-cycle periods. Nothing here says when a phase will end.

It does not know anything about you. The reading is identical for a twenty-five year old saving for a first home and a retiree drawing income, and those two people should almost certainly do different things with the same information. Suitability is not something a market-wide indicator can speak to.

It is not complete. Five categories built from free public data cannot capture everything that moves markets, and the proxies for profits and inventories are genuine compromises rather than best-in-class measurements. CORVIX publishes what feeds each category precisely so that you can judge how much weight the answer deserves.

Common questions

Does a late-cycle reading mean I should sell?

No. Late cycle is a description of conditions, not an instruction, and late-cycle periods have historically lasted a long time and produced strong returns before they ended. What the reading is useful for is checking whether your positioning still matches your intent, rather than triggering a decision on its own.

Why does the phase sometimes disagree with the market?

Because they measure different things. The cycle phase is built from economic data, while markets price expectations, and markets routinely turn several months before the data confirms the turn. A disagreement between the two is normal and is often the most interesting thing on the page.

How is the business cycle phase actually calculated?

Five categories, namely growth, credit, corporate profits, policy stance and inventories, are each scored from published data and fed into two axes: the level of activity and its momentum. Crossing those two axes produces the four phases. Growth carries the heaviest weight and inventories the lightest, the weights are fixed rather than fitted after the fact, and there is no discretionary override at any point. The full breakdown for each region, including which data source drove each category, is in the app.

Why does CORVIX classify each region separately instead of giving one global reading?

Because regions genuinely sit in different phases at the same time, and a single United States figure stretched over the rest of the world hides that. Central bank cycles are not synchronised, credit cycles are largely domestic, and a commodity move that hurts an importing economy helps an exporting one. CORVIX classifies all seven regions it covers on the same rubric, so a portfolio with international exposure can be checked against the cycle each holding is actually in.

What is the difference between the cycle phase and the macro quad?

They answer different questions on different clocks. The cycle phase asks where the economy sits in its expansion or contraction, and it is built from economic data that is released weekly and monthly, so it moves slowly. The quad asks only what growth and inflation are doing right now, accelerating or decelerating, and it moves considerably faster. Reading both together is more informative than either alone, and they can disagree without either being broken.

Can the phase be wrong?

Yes, in two distinct ways. It can be wrong because the underlying data was revised after publication, which happens routinely with GDP and employment figures. And it can be right about the economy while being unhelpful about markets, because markets price expectations and routinely turn several months before the data confirms the turn. Neither is a bug in the classification. Both are reasons to treat the reading as one input rather than a decision rule.

How far behind the data is the reading?

It depends on the category. Purchasing managers' indices arrive within days of month end, GDP arrives quarterly and is revised more than once, and credit aggregates typically lag by a month or two. The phase therefore describes the recent past rather than this morning, which is inherent to economic data and not a limitation CORVIX can engineer away. The quad and the correction risk reading are built from faster-moving market data and update accordingly.

What does a low confidence reading mean?

Every category is weighted, and the weights are renormalised over whatever data actually arrived. Confidence is the proportion of the possible weight that was genuinely present. A region missing several inputs still gets a phase, computed from what is available, but it is marked low confidence so the label is not read as though it were as well supported as a complete one. When a region has no growth data at all, CORVIX holds the previous phase and marks it rather than guessing from momentum alone.

Is this investment advice?

No. This page is informational only. It describes market conditions, does not know your circumstances or time horizon, and is not a recommendation to buy or sell anything.

Which sectors have historically done well in each phase?

Broadly: Early has favoured Consumer Discretionary, Financials, Industrials and Real Estate as credit reopens and deferred demand is spent; Mid has been the flattest phase, tilting mildly toward Technology and Communication Services; Late has favoured Energy, Materials, Consumer Staples and Healthcare as input costs rise and policy tightens; and Recession has concentrated into Consumer Staples, Utilities and Healthcare, whose revenue barely moves with the economy. CORVIX publishes the exact tilt table it applies rather than describing it in general terms. These are tendencies across many cycles rather than rules, and CORVIX combines the tilt with live valuation, momentum, trend and crowding signals rather than acting on it alone.

Is a Recession reading the same as an official recession?

No, and the two can disagree for a long time. Official recession dating in the United States is done retrospectively by the NBER, which waits for data to settle and revisions to land before announcing anything, so its calls have historically arrived many months after the period they describe. A Recession reading here is a real-time classification from currently published data. It can appear well before an official call, and it can appear when no official recession is ever declared.

Can I check what this framework would have said in the past?

Yes, and that is deliberate. Every input is point-in-time, meaning the reading for any past month uses only data that had actually been published by then, and the weights are fixed rather than fitted to historical results. The backtester in the app replays the same rules across history so the framework can be judged on what it said at the time rather than on what it would say with hindsight. Replaying a model with revised data makes almost any model look better than it was, which is exactly what point-in-time construction is there to prevent.

Does the cycle phase apply to individual stocks?

Not directly. The phase describes an economy, and the sector tilts derived from it describe broad groups of companies whose revenue responds to the cycle in similar ways. An individual company can diverge from its sector completely on the strength of its own products, balance sheet or management. The phase is best understood as a statement about the environment a company is operating in, not a statement about the company.