The dollar index is lying to you

A frozen relic from 1973

I got a comment from a subscriber some time ago that made me stop and think. (if it was you, please reply in the comments/DM so I can attribute you)

“Perhaps someone should have a closer look into the USD index calculation method. My nose tells me it’s useless and graphs with it are distracting.”

---

My immediate reaction? Nah, can’t be that bad. The DXY is everywhere. Every financial site quotes it. Every metals analyst references it. Hell, I reference it constantly. Surely if it was fundamentally broken, someone would have fixed it by now.

Right?

So I went digging. Pulled the ICE methodology documents. Read the Fed’s papers on trade-weighted indices. Traced the historical evolution of the basket composition.

And you know what I found?

His intuition was spot on! The DXY isn’t just useless for modern analysis. It’s a perfectly preserved fossil from 1973, maintained by a private company with every financial incentive to keep it that way.

The Federal Reserve figured this out back in 1998 and basically said “fine, we’ll make our own index”. They did. It tells a completely different story. And almost nobody uses it.

Let me show you what I mean.

The DXY launched in March 1973, right after Bretton Woods collapsed. The dollar was no longer pegged to gold, currencies were floating freely, and someone needed to figure out what the dollar was actually worth against its trading partners.

Fair enough. Reasonable goal.

Here’s the basket they created, weighted by 1973 trade patterns:

Euro: 57.6%
Japanese Yen: 13.6%
British Pound: 11.9%
Canadian Dollar: 9.1%
Swedish Krona: 4.2%
Swiss Franc: 3.6%

Wait, you’re thinking. The euro didn’t exist in 1973.

Correct. Bear with me.

The original basket had ten currencies. West German mark (20.8%), French franc (13.1%), Italian lira (9.0%), Dutch guilder (8.3%), Belgian franc (6.4%), British pound (11.9%), Japanese yen (13.6%), Canadian dollar (9.1%), Swedish krona (4.2%), and Swiss franc (3.6%).

And here’s where it gets fun.

The European currencies totaled 57.6% of the index.

When the euro launched in January 1999, those five European currencies got replaced by a single EUR entry at the exact same combined weight. Everything else stayed identical.

One change. That’s it. In 52 years, the DXY basket composition changed exactly once, and that was a mechanical substitution to accommodate a new currency. Not a rebalancing. Not an update to reflect shifting trade patterns.

One. Single. Change.

Let’s talk about who the United States actually trades with.

Top bilateral trading partners by total trade volume (January-October 2025, US Census Bureau):

1. Mexico: $731.2 billion

2. Canada: $606.7 billion

3. China: $357.2 billion (down 26.7% due to tariff war)

4. Taiwan: $201.1 billion

5. Germany: $196.4 billion

6. Japan: $190.7 billion

7. United Kingdom: $133.5 billion

(Note: The European Union as a bloc totaled $883.3 billion, making it the largest if counted as a single entity)

Now let’s see which of those are in the DXY basket.

Canada? Yes, at 9.1%.
Japan? Yes, at 13.6%.
Germany? Technically yes, buried inside the 57.6% euro blob.
United Kingdom? Yes, at 11.9%.

And the others?

Not there. Mexico isn’t in the basket. China isn’t in the basket. South Korea isn’t in the basket. The top three US trading partners by volume, and two of them don’t exist in the index that’s supposedly measuring dollar strength against trade partners.

But you know what is in there?

Sweden, at 4.2%.

Nothing against Sweden. Lovely country. IKEA, ABBA, quality meatballs. But Sweden-US trade is a rounding error compared to China-US or Mexico-US trade.

Switzerland gets 3.6%. Again, nice place. Good chocolate. But we’re talking about an index weighted by 1973 trade patterns when Swiss banking secrecy actually meant something and the yuan was barely convertible.

The world has changed a “little” bit since then.

1997 Federal Reserve Board

1998. The Federal Reserve created their own dollar index. Not to compete with ICE. Not to undermine the DXY. They created it because the DXY had already become useless for actual economic analysis.

The Fed’s Broad Dollar Index includes 26+ currencies (TradingView: DTWEXBGS). It rebalances annually based on current bilateral trade data. It includes both goods and services trade (DXY only ever used goods). And it has a competitiveness weighting factor for how much foreign producers compete with US products in third markets.

Let’s compare the top weightings:

DXY (frozen 1973 patterns):

• Euro: 57.6%

• Yen: 13.6%

• Pound: 11.9%

• Canadian Dollar: 9.1%

• China: 0%

• Mexico: 0%

Fed Broad Index (March 2025, updated annually):

• Euro Area: 21.1%

• Japan: 5.3%

• United Kingdom: 5.3%

• Canada: 13.8%

• China: 11.2% (tariff war impact - was 16% before)

• Mexico: 14.7%

• …

See the difference?

In the Fed's index, China and Mexico account for ~26% of the weighting. In the DXY, they account for exactly zero percent. The euro drops from 57.6% to 21.1%. Suddenly Japan’s weight gets cut by more than half.

The Fed’s index captures this.
The DXY doesn’t even know these countries exist.

This isn’t even a minor methodological dispute. These are completely different pictures of what the dollar is doing.

The DXY has a correlation of roughly 0.85 with EUR/USD. Inverted.

You could literally just chart EUR/USD backwards and get almost the same information. The euro is 57.6% of the index. When the euro falls, the DXY rises. When the euro rises, the DXY falls.

It’s basically the Anti-Euro Index.

And this creates some truly bizarre distortions.

Take 2025. The DXY fell about 10% year-to-date. Big move. “Dollar weakness”, right?

Except USD/CNY stayed remarkably stable around 7.30. The yuan didn’t strengthen against the dollar. It actually weakened against other currencies while holding steady against the dollar. Chinese authorities were explicitly managing that relationship.

But because China isn’t in the basket, the DXY doesn’t capture any of that. It sees the euro strengthening, registers “dollar weakness”, and ignores the fact that the dollar’s relationship with its largest trading partner barely moved.

The math is perfect. The calculation is transparent. The result is meaningless.

Why hasn’t anyone fixed this?? Well… No1 has any incentive to fix it. Neither does ICE.

The U.S. Dollar Index is a registered trademark owned by Intercontinental Exchange. It’s not a public utility. It’s a private company’s futures contract that trades 21 hours a day with $1,000 of value per index point.

Changing the methodology would disrupt decades of historical data. It would invalidate trading strategies built on 50 years of continuity. It would create legal and regulatory headaches around contract specifications.

And for what? The Fed already made a better index for people who actually want economic analysis. Everyone else just wants a liquid, tradeable dollar proxy, and the DXY works fine for that.

From ICE’s FAQ:

“There are no regularly scheduled adjustments or rebalancings of the ICE U.S. Dollar Index. The Index was adjusted once when the euro was introduced as the common currency for the European Union bloc of countries. ICE Futures U.S. monitors the index methodology to ensure that it properly reflects the covered currencies and the FX market in general.”

Translation: We’re watching it. We’re not changing it.

I searched for any evidence that ICE even considered adding China to the basket. Nothing. Not a word. China became the world’s largest exporter, the second-largest economy, a top-three US trading partner, and ICE’s response was apparently to shrug and keep selling futures contracts.

Why would they change? The contract is profitable. It’s established. It works for its intended purpose, which is providing a tradeable derivative, not measuring economic reality.

The perverse incentive isn’t manipulation. It’s inertia.

Now, is this something like the CPI being gamed by those controlling it?

Not quite… But the end result is similar.

CPI gets “adjusted” constantly. Hedonic quality adjustments. Substitution effects. Geometric weighting instead of arithmetic. The methodology changes to achieve a desired outcome - namely, to report a lower inflation than people actually experience.

The DXY doesn’t get adjusted at all. That’s the problem here.

With CPI, the game is active. They’re constantly tweaking the formula to minimize the reported number. With DXY, the game is passive. They’re freezing the formula while the world changes around it.

Both fail to measure what they claim to measure. One through constant tinkering, the other through militant stasis.

Pick your poison.

If you’re using the DXY to analyze gold and silver, you’re essentially tracking the dollar-euro relationship with some Japanese yen noise thrown in.

Asian demand? Not captured. The yuan has zero weight. Mexican mining output? Not captured. The peso has zero weight. Real dollar strength against emerging markets? Not captured. No EM currencies in the basket at all.

When you see analysis saying “DXY down, therefore metals should rally”, what that’s really saying is “EUR/USD up, therefore metals should rally”. Which might be true! The euro-dollar relationship does matter.

But it’s not the whole story. Not even close.

The Fed’s Broad Index fell about 10% in 2025 too, but the composition of that move was completely different. The euro strengthened, yes. But the yuan relationship stayed flat. The peso weakened. Emerging market currencies moved all over the place.

If you’re trying to understand physical demand dynamics in Asia, or production costs for Latin American miners, or real purchasing power for international buyers, the Fed’s index is going to tell you a lot more than DXY.

The DXY will tell you whether European money is flowing toward or away from dollar assets. That’s useful information. It’s just not the same as “dollar strength”.

I see analysis all the time showing DXY charts from 1973 to present, drawing trendlines, making long-term projections. And I thought this was a reasonable representation of reality. Now I know better. You do too.

Because what does that chart actually show?

1973-1998: Dollar strength vs actual major trading partners (flawed but functional) 1999-2010: Dollar strength vs euro-dominated basket (China already a major partner, not included) 2010-2026: Dollar strength vs an increasingly irrelevant basket (top three trade partners, two missing)

You’re comparing 1973 apples to 2026 oranges and calling it a trend.

In 1973, the Soviet Union still existed. China was pure communist and closed. Mexico just a backwater economy. The European Community? Nine members and a major US trade partner.

Compare that to the world today: China is the workshop of the world. Mexico is a top-three trade partner. The eurozone economy is important but not remotely as dominant as the 1973 patterns suggests.

Using a 50-year DXY chart assumes trade relationships are static. They’re not. They’ve transformed completely.

Ok, we’re all the wiser now. But what can we actually use in our analyses then?

For short-term euro-dollar dynamics: Use DXY. It’s liquid, it’s established, and it actually does track EUR/USD pretty well.

For everything else: Use the Fed’s Broad Dollar Index.

The Fed publishes three versions:

• Broad: All 26+ currencies (DTWEXBGS)

• Advanced Foreign Economies: vs the developed markets (DTWEXAFEGS)

• Emerging Market Economies: vs the EM currencies (DTWEXEMEGS)

You can find them on FRED (St. Louis Fed database). The series codes are DTWEXBGS for nominal broad index and RTWEXBGS for the real (inflation-adjusted) version.

They update daily. They rebalance annually. They include China at 11% and Mexico at 14%. They incorporate services trade, not just goods. They’re designed for economic analysis, not futures trading.

And they tell a completely different story about what the dollar is actually doing.

Conclusion

The DXY is a perfectly calculated relic.

It’s not a conspiracy. It’s not being actively manipulated. It’s just frozen in a world that no longer exists, maintained by a private company with every financial reason to leave it that way.

The DXY is useless for long-term analysis of actual dollar strength. It’s excellent for tracking the dollar-euro relationship and providing a liquid trading vehicle.

Those are not the same thing.

The Federal Reserve figured this out 26 years ago and built a better mousetrap. Most people still use the old one because it’s familiar and tradeable and nobody wants to explain to their readers why they’re switching metrics mid-analysis.

But if you’re trying to understand real dollar dynamics - especially in metals markets where Asian demand matters enormously - you’re handicapping yourself by relying on an index that thinks China doesn’t exist and Sweden is a major trade partner.

The math is perfect.

The inputs are just garbage.