MacroLens™ Gold Terminal

GII indicator centre

GII combines macroeconomic and capital-flow layers on a −100 to +100 scale. Its score is a reference, not a probability that gold will rise.

A component can change the gold allocation adjustment only after passing both the development and locked-data stages. Passing on locked data alone is not validation.

GII v2 signals: for M15, H1, H4 and D1 a concrete action (buy, sell or stand aside) is shown only under pre-registered rules: one side at least 60%, and measured out-of-sample accuracy at least 3 points above the base rate with at least 50 cases. Each action shows that accuracy, the sample size and the period. Entry zone, stop-loss and take-profit levels are for signed-in members only.

The 1 week to 6 month columns are statistics from the 20 most similar past contexts. They are not walk-forward tested and never trigger an action; only the 1 month column is pre-registered and is being tracked forward. Reference only, not personal investment advice.

How to read the “Gold potential” line, for example “after 1 month: up 65% — down 35%”: among the 20 past situations most similar to today, gold was higher one month later 13 times and lower 7 times, shown as 20 green and red dots. The line compares this with the normal rate, the share of all past one-month periods in which gold rose, and labels the tilt: under 5 points about normal, 5–10 slightly, 10–20 clearly, 20 or more strongly tilted up or down. These are counts of what happened, not a probability or forecast. The 1 week, 3 month and 6 month rows use the same wording, each with its own normal rate; their count column reads “up/down (out of 20)”, for example 13/7 or 10/10.

Blind spots and certainty (supplementary analysis registered 21 Nov 2026, not part of the v2 pre-registration; the engine was not changed). Out-of-sample accuracy of buy/sell actions versus the base rate: M15 56.3% vs 47.3% (n = 1,010, 2021-01 to 2026-11); H1 61.2% vs 50.2% (n = 412, 2021-01 to 2026-11); H4 53.1% vs 52.5% (n = 260, 2020-03 to 2026-11), practically no better than the base rate; D1 60.4% vs 56.2% (n = 642, 2007-04 to 2026-11). Entering one full bar late lowers M15 to 51.0% and D1 to 58.3%. Within ±2 hours of FOMC, CPI or payrolls each intraday frame has only 3 to 12 actions, too few to judge, so GII stands aside in those windows.

Latest public figures

Locked-data tests passed
9/254 (reference, not yet validated) (Public preregistered test table)
Gold allocation adjustment
0% (MacroLens conclusion rule)
GII score (−100 to +100)
-10 (MacroLens GII, 2026-11-20; not a probability)

Questions and answers

Is the GII score a probability that gold will rise?

No. GII is a composite reference score. A probability would require calibration and validation.

What does “reference, not yet validated” mean?

The result has not passed the required development and locked-data stages together. A locked-data pass alone does not confirm it.

Is this personal investment advice?

No. MacroLens supports decisions using dated data and public rules. It does not provide personal investment advice or guarantee returns.

What happens when a figure is unavailable?

Missing public data is shown as —. We do not fill gaps with invented numbers.

Open the interactive tool (Vietnamese)

Not personal investment advice. Each figure gives its source and period; unavailable data is shown as —.