Skip to main content
Ramin Khavarzadeh DATA · ANALYSIS · PERSPECTIVE
Lab INDEX · DEMAND PULSE

Demand Pulse Experimental

How strong is the appetite for large durable goods? — جستجوی خرید خودرو، خرید خانه، اجاره خانه، و میانگین سبد پنج کالای بادوام.

7 -4 Score vs. prior week

Percentile rank for week starting 18 Oct among 47 recent weeks; composite z -0.66. 2 Component up and 1 component down.

Components

ComponentDirectionRaw valuePrior weekz This weekz Prior weekChange
Search for “buy car”+45.0047.00-0.47-0.31Down
Search for “buy house”+22.0022.00-0.61-0.61—
Search for “rent house”+24.0023.00-0.50-0.71Up
goods_basket+37.0035.40-1.34-1.83Up

Each component is standardized (z) to the mean and standard deviation of this window, with a sign so “+” always means more of what the index measures. Index = simple mean of z’s, smoothed with a 3-week moving average, then percentile-ranked in the same window. Because mean and SD come from the whole window, past weeks are redefined as new weeks arrive — an in-sample view, not a signal that was available in real time; the lab’s own log below is what was actually shown each day.

Does this index say anything about the weeks ahead?

TargetBest lead (weeks)rpOut-of-sample direction accuracyResult
بازده هفتگی طلا2-0.230.13453٪ (n=17)descriptive

Test: correlation of the index composite with subsequent weekly returns (leads of 1–4 weeks) and direction accuracy on the final 40% of data that did not choose the lead. “Descriptive” means the index summarizes the present but showed no predictive power in this test — and is labeled that way, not as a forecast.

Sources
Google Trends (IR)، tgju
Cadence
Weekly; search refreshes every 12 hours and markets every 6 hours; index recomputed hourly
Window
47 weeks
Lab’s own recorded history
53 Day
Last calculation
2026-11-02T20:51

Limits: Google search interest is relative and rescaled on every fetch, so past weeks may shift slightly; low-volume queries have integer resolution; the window is one year, so “strongest week of the year” only means within that year. None of these indices are forecasts. Full methodology