Independent strategy research
The intermarket “rules” your charts still assume — re-tested
Short version: two of the three famous ones don't hold the way the textbook says. Gold's classic −0.8 link to real yields flipped to +0.3 in 2023–2026 while gold rallied 100%+ and real yields rose about a full point. The oil–stock line is an average of two opposite regimes. And the dollar–commodities–bonds–stocks “sequence” is folklore.
Trading education is full of confident intermarket “rules.” Gold moves opposite real yields. Oil and stocks move together. The dollar leads commodities, which lead bonds, which lead stocks. They get drawn as if they were laws of physics. We re-tested the famous ones on current data. Here is what held and what didn't.
1. Gold vs real yields — the classic relationship broke
For decades gold moved strongly opposite real yields. Real rates up, gold down — the textbook level correlation sat near −0.8, and a generation of traders learned the reflex: short gold when real rates rise.
In 2023–2026 it flipped. The level correlation went from about −0.4 to +0.3. Gold rallied 100%+ while real 10-year yields rose about a full percentage point. The old reflex would have had you short into a doubling. Central-bank buying and de-dollarisation overwhelmed the rate channel, and the sign inverted. Any gold-versus-rates model has to carry a post-2022 break or it is describing a market that no longer exists.
2. Oil vs stocks — the sign depends on the shock, not the price
There is no stable oil-stock correlation; its sign is set by what moved oil in the first place. Use copper as a demand proxy. When oil moves with copper — both rising on stronger demand — the oil-stock correlation is strongly positive, around +0.42. When oil moves against copper, an oil-specific or supply shock, it flips negative, around −0.10.
Blend the two regimes and you get an unconditional +0.25 that describes neither. Regressing stocks on the raw oil price is misspecified. The tidy “oil-stock” line on a chart is the average of two opposite worlds, and averaging opposites is how you get a number that is always slightly wrong.
3. The dollar → commodities → bonds → stocks sequence — folklore
The classic lead chain has plenty of chart appeal and no peer-reviewed out-of-sample support. We tested it the honest way: as a formal directed lead structure on a walk-forward basis, asking whether any link actually leads the next in a way you could trade ahead of. It doesn't. No link led the next tradeably. The co-movement is contemporaneous — these markets move together, they don't queue up. A sequence you can only see in hindsight is not a sequence you can trade.
What this means if you trade off intermarket signals
These relationships are real and worth understanding — for risk, correlation and diversification. That is not the same as a directional signal. The moment you treat one as a stable rule that says buy this because that moved, you've assumed a constant that the last three years just disproved. Gold's inversion is the clean example: the correlation didn't weaken, it changed sign. A hedge built on the old number would have added risk, not removed it.
This is the same pattern we find across the whole audit. A relationship that looks bankable on a chart is usually a regime average, and regimes move. That's why intermarket analysis earns a “conditional” verdict from us, not a “yes” and not a flat “no.”
How we tested this
We re-ran each relationship on current-window data with explicit regime splits, rather than reading a single correlation off a full-history chart. Oil's shock type is proxied by copper co-movement, in the spirit of the Kilian-Park decomposition that separates demand shocks from supply shocks. The lead chain was tested as an out-of-sample directed increment, not a chart overlay you eyeball after the fact. Same discipline we apply to strategies and indicators: futures from 13 years of CME data, tick-level FX with real bid/ask, liquidity-aware stocks, crypto spot and perps. Over 2,700 scripts have been through the pipeline the same way.
Research and education, not financial advice. No signals, no return promises. These are relationship and regime findings, not tradeable directional edges — and relationships can shift again. Independent, and not affiliated with TradingView.
The relationships are public. The verdicts aren’t.
This page is the free version: aggregate findings, no names. What it doesn’t give you is which published intermarket and macro strategies we tested, and the exact after-cost verdict on each one. That is The No List — every strategy and indicator we audited, named, with the reason it lived or died.
Get The No List → or see the free research in our Discord →FAQ
Does gold still move opposite real yields?
Not in the current regime. The textbook level correlation was near −0.8; in 2023–2026 it went from about −0.4 to +0.3. Gold rallied 100%+ while real 10-year yields rose roughly a full percentage point. The old “short gold when rates rise” reflex inverted.
Is the oil and stock market correlation positive or negative?
Both, depending on the shock. When oil rises on demand (moving with copper), the oil-stock correlation is about +0.42. When it's an oil-specific or supply shock (oil against copper), it's about −0.10. The blended, unconditional number is +0.25 and describes neither regime.
Does the dollar lead commodities, bonds and stocks in sequence?
Not tradeably. Tested as a formal directed lead structure out of sample, no link led the next in a way you could act on ahead of time. The co-movement is contemporaneous, not a queue you can front-run.
So is intermarket analysis useless?
No — it's useful for risk, correlation and diversification. It stops being reliable the moment you treat a relationship as a fixed directional signal, because the relationships are regime-dependent and regimes move. We rate it conditional, not yes or no.