Education · Free
Bonds, Gold, Oil, and FX: Check the Relationship Before Using It
Separate conversion arithmetic, economic channels and sample-dependent correlations before using familiar bond, gold, oil and currency relationships.
Image credit
Deliberately constructed four-point samples, not empirical intermarket relationships.
FavAcademy · Original FavAcademy educational illustration; public use authorized
Public summary
Bonds, gold, oil and currencies can move together or apart as the underlying shock changes. Two deliberately constructed return samples show opposite correlations, illustrating why a familiar relationship needs matched data, a defined window and a conditional explanation.
Begin with a relationship, not a slogan
Intermarket commentary often compresses a conditional idea into a slogan: gold rises when the dollar falls, oil supports an exporting currency, or higher bond yields strengthen a currency. Each story may describe a plausible channel in some circumstances. None is a complete rule for every period, instrument or shock. The first task is to state exactly which prices or returns are being compared and why a link might exist.
A currency pair is relative, while commodities and bonds have their own quotation and contract conventions. A dollar-denominated gold price is not the same exposure as gold measured in another currency. A bond’s price is not its yield. An oil future has a maturity and can behave differently from a cash benchmark. Without matching definitions, apparent intermarket agreement can be a comparison of unlike quantities.
Separate arithmetic translation from an economic claim
Suppose a fictional asset costs 2,000 dollars per unit and the conversion rate is 0.90 home-currency units per dollar. Its home-currency value is 1,800. If the dollar price stays at 2,000 while the conversion rate rises to 0.95, the home-currency value becomes 1,900. That increase follows from conversion arithmetic; it does not require the asset’s dollar price to rise or an independent change in demand for the asset.
Conversely, simultaneous changes in the asset price and exchange rate can offset or reinforce one another. State the investor’s measurement currency before describing the relationship. A chart of dollar gold and a chart of home-currency gold need not tell the same story. The translation effect is a useful starting distinction because it can explain part of a move without pretending to identify its underlying cause.
Read bond yields through the shock
For a conventional fixed-rate bond, a rise in required market yield generally corresponds to a lower price, other things equal. Currency interpretation is a further step. A higher yield associated with stronger expected policy rates can carry different information from a higher yield associated with inflation concerns, credit risk or reduced demand for the bond. The same observed yield direction need not imply the same currency response.
Maturity matters too. A short-dated yield and a long-dated yield contain different mixtures of expected policy and other compensation. Comparing one country’s long yield with another’s overnight policy rate mixes horizons and risks. A yield differential alone does not establish a currency forecast or a profitable carry opportunity. Keep the instrument, maturity and proposed mechanism visible instead of turning “yields up” into an automatic conclusion.
Treat commodity channels as conditional
Oil prices can affect import bills, export receipts, inflation and policy expectations, with different implications for different economies. EIA discusses several channels linking oil and the dollar and explicitly notes that the relationship has become more complicated, sometimes showing positive correlation. That is a direct reason to resist a permanently inverse oil–dollar rule. The cause of the oil move and the economic exposure matter.
Gold narratives likewise need a specified channel, such as currency translation, financing conditions or demand during stress. Naming a possible channel does not quantify its strength or guarantee that it dominates every other influence. A common global shock can move both gold and a currency without one directly causing the other. Avoid assigning participant motives solely from the fact that two charts changed together.
Work a deliberately unstable correlation example
Take four fictional paired returns. Series X is 1%, 2%, negative 1% and negative 2%. In window A, series Y is negative 1%, negative 2%, 1% and 2%. Both means are zero, and Y is exactly the negative of X, so their Pearson correlation is negative one. In a separate window B, let Y equal X instead. The correlation is positive one.
These samples are constructed to make the arithmetic transparent, not measured relationships between named assets. The illustration shows the two sets of paired observations with opposite slopes. Four points and perfect relationships are not realistic evidence of stable predictability. The lesson is that a reported coefficient belongs to its data and window; a familiar label cannot force it to retain the same sign in another sample.
Check the measurement before using the story
Use aligned timestamps, the same return horizon and appropriate currency units. Comparing price levels that share trends can give a misleading impression, while closing times from different markets can pair non-synchronous observations. Record missing data and avoid filling gaps with convenient values. Pearson correlation describes linear association, not causation, and a low coefficient does not rule out other forms of dependence.
If you test several assets, windows and lags, disclose that exploration and retain the alternatives. Selecting the strongest historical relationship after trying many combinations can make chance look like discovery. Evaluate a chosen specification on separate observations before assigning practical significance. Even a persistent relationship would still need execution, cost and risk analysis before it became part of an implementable trading strategy.
Keep the argument conditional
A useful intermarket explanation names the instruments, measurement currency, horizon, sample and proposed economic channel. It also states what evidence would weaken the story. This is more informative than collecting several charts that appear to confirm the same direction, especially when they share a common dollar quotation or underlying price inputs.
The fictional translation rises from 1,800 to 1,900 home units despite an unchanged dollar asset price. The two constructed return windows have correlations of negative one and positive one. Neither is a market forecast. Together they teach a disciplined distinction: arithmetic relationships can be exact, economic mechanisms can be plausible, and empirical associations can still change. Treat each claim at the level of evidence it actually has.
Continue learning with What DXY Measures and What It Leaves Out and Rate Differentials: Why the Higher-Yield Currency Can Still Fall.
Risk disclosure
Returns and asset prices are fictional. Correlations depend on the sample and do not prove causation, stable diversification or a profitable strategy.