Intermarket analysis: how markets move together
The answer
Intermarket analysis studies how bonds, equities, commodities and currencies move in relation to each other. The links are real and they have economic mechanisms behind them, mostly running through interest rates and inflation. What is not reliable is the sign of any given relationship, because the same 2 markets move together in one monetary regime and in opposite directions in another.
Why this costs you money
The specific loss is this: you believe you are diversified when you own 1 bet.
A person holds equities, a bond fund, some gold and a little exposure to a foreign currency. Four things. It looks like 4 decisions. In 2022 all 4 of those positions were driven by the same variable, which was the speed at which the American central bank raised interest rates. When 1 variable drives 4 positions, you own 1 position in 4 wrappers, and it will fall as 1 thing.
The second loss is smaller and more frequent. You learn a relationship as a rule and keep applying it after the rule has flipped. "Bond prices and share prices move together" was a good description of American markets for roughly 30 years before 1998. "Bond prices and share prices move in opposite directions" was a good description for roughly the next 20 years. Then in 2022 they fell together again. Somebody using any 1 of those 3 statements as a permanent law was wrong for a long stretch of their investing life.
The third loss is the most expensive and the least visible. You use a correlation as a hedge without checking that it holds in stress. Correlations between risky assets rise sharply during crises. The diversification you measured in calm conditions is measured on the wrong sample, and the protection disappears precisely on the days you needed it.
How it works
John Murphy set out the modern framework in Intermarket Technical Analysis in 1991 and expanded it in Intermarket Analysis in 2004. The idea is that 4 asset classes form a system, and that a turn in one often precedes a turn in another.
The 4 links, as usually taught
1. Currency and commodities. Most globally traded commodities are priced in United States dollars. A stronger dollar makes them more expensive for buyers using other currencies, which tends to reduce demand and push prices down. So a rising dollar is usually associated with falling commodity prices.
2. Commodities and bonds. Rising commodity prices feed into consumer prices. Higher expected inflation makes a fixed stream of future interest payments worth less, so bond prices fall and bond yields rise.
3. Bonds and equities. Bond yields set the rate at which future company profits are discounted. In the framework as usually taught, bond prices turn before share prices, so a sustained fall in bond prices is an early warning for equities.
4. Equities and commodities. Equities are usually taught to lead the economy and commodities to follow it, so commodities tend to peak after equities.
Why the chain exists at all
Read the chain again and notice that every link goes through the same 2 variables: inflation and the interest rate. That is the mechanism. It is not a chart property, and it is why intermarket analysis is more defensible than anything in cluster 18.
Higher expected inflation raises nominal interest rates. Higher interest rates lower the present value of distant cash flows, which hurts long-duration assets: long bonds first, then the shares of companies whose profits are far in the future. A stronger currency lowers the domestic price of imported goods, which lowers measured inflation, which lowers the pressure on rates.
Everything in the framework is a consequence of that. Once you see this, the framework stops being 4 rules to remember and becomes 1 idea applied 4 times.
Why the signs flip
Here is the part that decides whether you use this well or badly.
A correlation is not a mechanism. It is the result of whichever shock happened to be dominant during the period you measured.
Take shares and bonds. If the dominant news is about growth, they move in opposite directions. Weak growth is bad for profits, so shares fall, and weak growth means lower rates, so bond prices rise. If the dominant news is about inflation, they move together. Higher inflation raises rates, which hurts bond prices, and the same higher rates hurt share valuations.
So the stock-bond correlation is not a fact about stocks and bonds. It is a readout of which kind of shock the market is currently reacting to. Antti Ilmanen made this argument in the Journal of Fixed Income in 2003, and John Campbell, Carolin Pflueger and Luis Viceira developed a formal version in the Journal of Political Economy in 2020.
The same applies to oil and shares. Lutz Kilian showed in the American Economic Review in 2009 that oil price increases have very different effects depending on their cause. An increase caused by a supply disruption harms importing economies. An increase caused by strong global demand comes with a strengthening economy and is often associated with rising share prices. Identical price move, opposite meaning, and no chart can tell you which it was.
The 4 questions that replace the 4 rules
Rather than memorising signs, ask what is driving the current period. There are broadly 4 kinds of dominant shock, and each produces a different set of signs.
| Dominant driver | Shares | Bond prices | Commodities | Safe currencies |
|---|---|---|---|---|
| Growth improving | Up | Down | Up | Weaker |
| Growth deteriorating | Down | Up | Down | Stronger |
| Inflation rising | Down | Down | Up | Depends on policy |
| Risk event or panic | Down | Up, usually | Down | Stronger |
Read across the rows. Notice that "shares down" appears in 3 of the 4 rows, and bond prices behave differently in each of them. That is why "bonds hedge shares" is right most of the time and catastrophically wrong in an inflation shock.
The correct first question is not what the relationship is. It is which row you are in. If you can name the dominant driver, the signs follow. If you cannot, no amount of chart overlay will supply it.
Lead and lag, honestly
The framework claims some markets lead others. Bonds are said to lead equities. Commodities are said to lag.
Treat these claims carefully. A lead that shows up in monthly data over decades is not the same as a signal you can act on this quarter, because the lead time varies enormously and is only measurable after the fact. A relationship that leads by "somewhere between 2 and 18 months" is a description of history, not a timing tool.
The defensible version is narrower and still useful. A large, sustained move in the bond market usually reflects a change in the expected path of interest rates, and that change will eventually be reflected in equity valuations. That is worth knowing. It is not a signal with a date attached.
What it tells you, and what it does not
Intermarket analysis tells you what your portfolio is actually a bet on. That is its most valuable use and it has nothing to do with forecasting. When you list your positions and identify the driver behind each, you often find that 6 positions share 2 drivers. This is a measurement of concentration you cannot get any other way.
It tells you when a market is disagreeing with another market. If share prices are making new highs while the bond market is pricing a sharp fall in rates, 2 markets are forecasting different things. One of them is wrong. Knowing which is not available, but knowing that the disagreement exists changes how large a position you should hold.
It does not give you the sign of a relationship in advance. You have to determine which regime you are in, and regime identification is only certain in hindsight.
It does not give you a stable number. Correlations computed over 1 year, 5 years and 20 years on the same 2 assets routinely have different signs.
It does not survive a crisis. Francois Longin and Bruno Solnik showed in the Journal of Finance in 2001 that correlations between international equity markets rise in the tails, and that the rise is concentrated in market declines rather than in market rises. Diversification measured on ordinary days overstates the diversification you will have on the worst days.
The decision rule
Before you treat 2 markets as related, name the mechanism and name the regime.
If you can state the economic channel linking them — through inflation, through interest rates, through input costs, through a currency in which one is priced — and you can name which of the 4 dominant drivers is currently operating, then use the relationship as context.
Unless the relationship is one you observed on a chart without being able to say why it exists. In that case it is a correlation found by looking, and it will end without warning.
And whatever the relationship says, size positions as though the correlation may go to 1 in a crisis. It usually does.
The practical version: do not ask what 2 markets did together. Ask what would have to be true for them to keep doing it.
Try this now
Ten minutes, on your own charts, testing a relationship you personally believe in.
- Write down 1 intermarket relationship you actually rely on. Examples people commonly hold: a weaker rupee is bad for Indian shares; higher oil prices are bad for Indian shares; a stronger dollar is bad for gold; rising bond yields are bad for technology shares. Write your own, in 1 sentence, before you look at anything.
- Open a chart platform that allows 2 series in 1 window. Most free platforms allow a comparison overlay. Put both series on 1 chart over the last 12 months.
- Now count instead of looking. Take the last 12 month-end dates. For each month, mark whether series A rose or fell, and whether series B rose or fell.
- Count how many of the 12 months moved in the direction your relationship predicts. Write that number.
- Extend the same count over the previous 12 months as well, so you have 2 separate 12-month counts.
- Compare the 2 counts.
What you should see. If your relationship is strong and stable, you will see something like 9 or 10 out of 12 in both periods. Most people do not see that. The common result is roughly 7 out of 12 in one period and roughly 6 out of 12 in the other, and 6 out of 12 is what a coin produces.
The comparison between the 2 periods is the real finding. A relationship that scored 10 in the first year and 5 in the second has not weakened. It has changed sign, and that is what regime change looks like from the inside.
Then do the last step, which takes 1 minute and is the most useful. Write down the mechanism. In 1 sentence, say why A should affect B. If you cannot write the sentence, you have been holding a correlation, not a relationship, and the count you just did is the only evidence you have for it.
Three real cases
1. Calendar year 2022, United States — the year the standard hedge stopped working Through 2022 the Federal Reserve raised its policy rate rapidly in response to inflation. American share prices fell and American bond prices fell at the same time, across the whole year. Long-dated Treasury bonds had one of their worst years on record, and a conventional portfolio of 60% shares and 40% bonds recorded one of its worst calendar years since the 1930s.
Everything in this outcome is explained by the table above. The dominant driver was inflation, not growth, and in an inflation shock shares and bonds fall together. Anybody whose risk model was estimated on data from 1998 to 2020, when the dominant driver was growth, had a model that could not produce this outcome at all.
2. The taper tantrum, India, May to September 2013 — one shock arriving through 3 markets at once On 22 May 2013 the chairman of the United States Federal Reserve indicated that the pace of bond purchases might be reduced. Capital flowed out of emerging markets over the following months. The Indian rupee fell to a record low against the dollar in late August 2013, reaching a level near 68 to the dollar . Indian government bond yields rose sharply. Indian equities fell. The Reserve Bank of India tightened liquidity and later launched a special swap window for foreign currency deposits from non-resident Indians .
The case matters because 3 Indian markets — currency, bonds and equities — moved together in response to a single piece of news from another country. An Indian investor holding all 3 held 1 exposure. The intermarket framework does not predict the event. It tells you, in advance, which of your positions will move as 1 when it happens, and that is a usable output.
3. March 2020, both markets — when everything correlates In the second and third weeks of March 2020 almost every asset fell at the same time. Equities fell in every major market. Corporate bonds fell. Gold, which is widely held as a hedge, fell across several sessions in mid-March before recovering strongly later in the year. Even United States Treasury markets showed severe strain, which led the Federal Reserve to intervene in the market for Treasury securities .
The mechanism is not mysterious. When investors face demands for cash, they sell what they can sell, not what they want to sell. That makes ordinary asset relationships irrelevant for a period.
This is the case to remember when you compute a correlation. The number you compute is dominated by ordinary days, because there are thousands of them. Your portfolio's survival is decided by the 10 worst days, when the number does not apply.
The question that resolves it
A novice asks: are these 2 markets correlated?
An expert asks: what would have to change for that correlation to reverse, and has it already started changing?
The second question has an answer you can look for. For shares and bonds, the answer is whether the market is currently reacting mainly to growth news or mainly to inflation news. You can observe that directly: watch what happens to share prices on the day an inflation figure is released. If shares fall on high inflation and rise on low inflation, you are in the inflation regime and bonds will not hedge your shares. If shares rise on strong growth data and fall on weak data, you are in the growth regime and bonds probably will.
That is a 30-second observation, repeatable every month, and it tells you more than a 20-year correlation coefficient.
What would make this wrong
This article says the links are real, mechanism-driven and unstable in sign. Each part can fail.
The mechanism claim would be wrong if asset prices stopped responding to interest rates and inflation. There is no serious argument that this is happening, and it is close to a definition of how discounting works.
The instability claim would be wrong if the sign of the major relationships turned out to be stable over long periods. It is not. The stock-bond correlation in the United States has changed sign more than once in the last 60 years, which is directly observable in the data and is not a matter of interpretation.
The usefulness claim is the one genuinely open to challenge. A critic can reasonably say: if the sign flips and you can only identify the regime after the fact, then intermarket analysis explains the past and forecasts nothing. That criticism is largely fair for forecasting. It is not fair for risk measurement, which is the use this article recommends. Knowing that 4 of your positions share 1 driver is valuable even if you have no idea which way the driver will move.
The honest limit. Almost all of the published intermarket work is descriptive. There is very little formal out-of-sample testing of intermarket trading rules, and the tests that exist face the same data-mining problem as every other technical rule: with 4 asset classes and many lag choices, a large number of relationships can be tested and the best one will look impressive by chance.
In India
The Indian version of this framework has 4 links that matter more than anything in the standard American teaching.
Crude oil. India imports the large majority of the crude oil it consumes . A rise in crude prices worsens the trade deficit, puts pressure on the rupee, raises domestic fuel prices and therefore inflation, and raises pressure on the Reserve Bank of India. That is a chain with 4 links and every one of them is real. Note Kilian's warning though: a crude rise caused by strong global demand arrives alongside stronger exports, and the net effect is smaller than the first link suggests.
Foreign portfolio flows. Foreign portfolio investors are a large marginal buyer and seller in Indian equities and, increasingly, in Indian government bonds. Their flows respond to global conditions, particularly to United States interest rates and to the dollar. When they sell, they sell shares and they sell rupees at the same time. This is why the rupee and Indian equities often fall together, even though a weaker currency helps exporters. The flow effect dominates the trade effect over short horizons.
The rupee is managed, not free. The Reserve Bank of India intervenes in the foreign exchange market and holds large reserves. That means the rupee's chart is partly a record of policy decisions rather than only of market pressure. A currency relationship that assumes a free float is being applied to something that is not one.
Gold has a domestic role the American framework does not include. Indian household demand for physical gold is very large, gold imports affect the current account, and import duty changes have moved the domestic price relative to the international price. Indian gold prices are also a currency position, because the domestic price is roughly the international price converted at the rupee rate. A rupee fall raises the rupee gold price even when the dollar gold price is unchanged.
Two more Indian details. Indian government bond yields are now more connected to global flows than they were, following India's inclusion in a major global bond index from 2024. And the Nifty 50 carries a very large weight in financial companies, which makes the Indian equity index unusually sensitive to interest rates directly, not only through discounting.
In the United States
The United States is where this framework was developed, and the data available is the best anywhere.
The dollar is the pricing currency for global commodities. That gives the dollar and commodity prices a direct arithmetic link that does not exist for any other currency. The dollar index and broad commodity indices have historically moved in opposite directions, though the strength of that relationship varies considerably by decade.
Treasury yields are the global discount rate. The 10-year Treasury yield is used as the reference rate for pricing risk assets around the world, including in India. This is the single most important number in the intermarket system, and its daily history is published free by the Federal Reserve in the H.15 release.
Policy is observable in unusual detail. The Federal Open Market Committee publishes statements, minutes 3 weeks later, and quarterly projections. Futures markets on the policy rate let you read the market's expected path directly. This means an American analyst can separate a growth shock from a policy shock far more cleanly than an Indian analyst can, because the expectations are explicitly priced.
The relationships have flipped repeatedly in the record. American equity and bond returns were positively correlated for much of the period from the 1960s to the late 1990s, negatively correlated for roughly 2 decades after that, and positively correlated again in 2022. Because the data is long, this is not a matter of opinion. It is the clearest available evidence that intermarket signs are regime-dependent.
Where they differ, and what that tells you
The mechanisms are the same. The direction of the causation is not.
One market sets the rate and the other receives it. United States interest rates affect Indian asset prices through capital flows and the exchange rate. Indian interest rates do not have a comparable effect on American asset prices. For an Indian investor this means the most important intermarket variable for your portfolio is set in another country, by a committee whose mandate does not include you.
The consequence is that Indian diversification is narrower than it looks. When American rates rise sharply, foreign flows out of India hit Indian equities, Indian bonds and the rupee at once. An Indian investor holding all 3 has 1 exposure. An American investor holding domestic shares and domestic bonds in the same episode has 2 exposures that at least have different mechanisms, even when they move the same way.
The currency is a separate position for an Indian investor and usually not for an American one. An American investing in American shares takes no currency risk. An Indian investing in a global fund takes an equity position and a rupee position at the same time, and over long periods the rupee has depreciated against the dollar. That means an Indian investor's returns from foreign assets contain a currency component that is not present in any American textbook example.
And the practical first step differs. An American reader should start by watching the relationship between share prices and interest rate expectations, because that is the dominant domestic channel. An Indian reader should start somewhere else: watch net foreign portfolio flows, published daily by the exchanges and by SEBI, because a single flow variable moves 3 of your markets simultaneously and is the closest thing India has to a master variable.
Carry this
- Every intermarket link runs through inflation or interest rates. One idea, applied 4 times.
- A correlation is not a mechanism. It is a record of which shock was dominant when you measured it.
- Name the driver first: growth improving, growth deteriorating, inflation rising, or panic. The signs follow from the row.
- The main use is not forecasting. It is finding out that 6 of your positions share 2 drivers.
- Correlations rise toward 1 in a crisis. Size positions as though they will.