THE QUESTIONS THIS TOOL ANSWERS
Did gold's return distribution differ—and did the observed state persist?
This is a conditional-history tool, not another correlation chart. It places every usable month into one of four transparent groups, shows every gold return behind the summary, and separately counts state paths between strictly adjacent months. Unequal and sparse samples remain visible instead of being compressed into one bullish or bearish score.
No usable source response
No value is substituted.No usable source response
No value is substituted.No usable source response
No value is substituted.No usable source response
No value is substituted.READING ORDER
Start with sample size, then distribution, then the path
A regime with three months is weaker descriptive evidence than one with twenty. No minimum converts the result into a forecast.
A wide difference can reveal the influence of one unusually large month. The strip plot keeps that outlier visible.
Matrix color answers how often a state moved to another state. It does not encode gold performance.
Verify driver moves, gold return and separate observation dates before drawing a conclusion.
METHOD BOUNDARY
Four states are built from consecutive month-end observations
For gold and each selected driver, GoldObserve keeps the last valid observation in each UTC calendar month. A month enters the analysis only when all three series also have a valid observation in the immediately preceding calendar month. Gold is measured as a percent return; real yield and inflation use percentage-point changes; the dollar uses percent change.
The transition matrix adds a second strict adjacency test: a classified destination month is connected only when the immediately preceding classified month is present. Its cell percentage is the share of transitions leaving that row state. Gold return is measured in the destination month, not the following month.
“Higher / flat” includes a zero monthly change. This binary direction rule is deliberately simple and reproducible, but it does not measure the size, surprise or publication timing of a macro move. CPI month labels are source observation periods, and daily series can have different final trading dates inside the same month.
WHAT THIS CANNOT ESTABLISH
Regime differences can reflect omitted variables, revisions and chance
- The tool uses latest available or latest revised macro observations, not an ALFRED point-in-time vintage.
- It does not control for policy expectations, liquidity, geopolitics, positioning, physical demand or starting valuation.
- Four states create sixteen possible transition cells, so many cells remain sparse even when the overall sample looks adequate.
- Observation-month alignment is not the same as aligning data by official release timestamp.
- Historical distributions and row shares are not price targets, transition forecasts or investment recommendations.
RELATED RESEARCH
Move between explanation, association and conditional history
GOLD MACRO REGIME FAQ
Questions about the classification
What is a gold macro regime?
Here it is a descriptive four-way grouping based on whether two selected series rose or stayed flat versus fell from the prior calendar month.
What does the transition matrix measure?
It counts how those four observed states changed between strictly adjacent classified months. A row percentage describes the historical sample leaving that row state; it is not a forecast probability.
Does the strongest group predict gold?
No. It only describes this selected, latest-revised historical sample.
Why use monthly observations?
Monthly grouping permits daily and monthly official series to be compared without pretending every source shares the same trading or release calendar.
Why are some months missing?
A month is omitted when gold or either selected driver lacks the current or immediately previous calendar-month observation. Missing values are not filled.