CalcPulse

Open data, cleaned weekly

Fourteen public macro and AI-ecosystem series, rebuilt every Monday with the week-over-week change, a 12-period z-score and an outlier flag already computed. The six-series sample below is free; the full pack is a $9 file with the whole table, twelve months of weekly history and the written briefing.

This week’s sample — 2026-W47

6 of 14 series · snapshot 2026-11-17T10:56:08Z

Metric (unit)LatestPrev (7d)Changez (12w)Flag
US 10-year Treasury yield% p.a. · observation 2026-11-135.1504.860+5.97%+2.71outlier
USD/KRW exchange rateKRW per USD · observation 2026-11-131,336.581,361.89-1.86%-0.89—
EUR/KRW exchange rateKRW per EUR · observation 2026-11-161,470.221,507.45-2.47%+0.15—
CBOE volatility index (VIX)index points · observation 2026-11-1319.3025.64-24.73%-2.24outlier
Nasdaq Compositeindex points · observation 2026-11-1624,509.7625,253.14-2.94%-1.06—
New repositories tagged topic:llmrepositories created per week · observation 2026-11-165,3865,015+7.40%+0.58—

The sample is a real slice of this week’s build, not a mock-up — the same twenty columns, six rows shorter. Download the sample CSV.

What the full pack holds

One file per week, same listing, same price. Every row keeps the public page it came from, so any figure can be traced back to its origin in one click.

weekly-macro-tech-diff-{week}.csv

14 rows × 20 columns: latest value, previous value, the exact dates both came from, absolute and percent change, a 4-week change, a 12-period z-score and an outlier flag.

weekly-macro-tech-series-latest.csv

The same series as tidy weekly history — long format, up to 52 periods, with the forming week marked partial so a chart does not compare half a week to a full one.

briefing-{week}.md

The written read: largest moves with the two values they were computed from, outliers, what did not move, and the method.

DATASET-README.md + LICENSE.txt

Column dictionary, refresh rules and terms. Use it inside your organisation; do not resell the pack itself.

CategorySeriesUnit
RatesUS effective federal funds rate% p.a.
RatesUS 10-year Treasury yield% p.a.
InflationUS CPI, all urban consumers (SA)index, 1982-84 = 100
FXUSD/KRW · EUR/KRW · JPY/KRW · CNY/KRW · EUR/USD · USD/JPYdaily cross rates
EquityNasdaq Compositeindex points
RiskCBOE volatility index (VIX)index points
AI ecosystemNew repositories tagged topic:llm · topic:ai-agents · topic:mcprepos / week

The full pack

$9, one week’s pack, no subscription

A new pack is built every Monday and replaces the file on the same listing. You get the CSV, the weekly history and the briefing; nothing else on this site is gated.

Get this week’s pack →

How the numbers are made

Sources. FRED, the Federal Reserve Bank of St. Louis download endpoint; the ECB euro reference rates retrieved through Frankfurter; and GitHub’s public search counts of repositories created each week. All free, all without an API key, so the pack costs nothing to produce and nothing to keep running.

Comparison. Daily series compare the latest print with the last print on or before the seven-day mark, and the actual gap is a column (window_days) — a holiday should not silently empty a row. Monthly series compare month over month. Weekly counts use rolling seven-day windows.

Outliers. The latest change is scored against the previous twelve changes of the same series, so a 0.2% move in VIX and a 0.2% move in a repo count are not treated as the same event. Below three prior changes the z-score is blank and the flag is FALSE rather than guessed.

The guard. If any series comes back empty — a changed schema, a blocked request, an empty body — the build exits non-zero and publishes nothing. The previous week’s pack stays the current one. A header-only CSV is never sold here.

What it is not. No forecast, no advice, no “intelligence”. GitHub totals are raw search counts and include bulk uploads and duplicates, which the briefing says out loud rather than hiding.

Where the numbers are used

The same figures drive the calculators: an exchange-rate move changes what an overseas API bill costs in won, and a rate move changes what a break-even month looks like.