Sugra Macro: a stale forecast looks exactly like a fresh one
Who published a figure, and on what date? This post follows the Sugra Macro video area by area: one live request and its answer for each of six areas, from the Federal Reserve's balance sheet to a composed country profile, pulled on 1 November 2026.
PublishedUpdated

In April 2025 the US unemployment rate was 4.2 percent. In September 2026 it was 4.2 percent again. Same agency, same series, same figure, seventeen months apart. The Sugra Macro video opens on that pair: an old unemployment figure looks exactly like a fresh one, and only its date, and who published it, tell them apart.
Forecasts make it worse, because they are meant to change. In October 2025 the IMF put German growth for 2027 at 1.473 percent. Its latest World Economic Outlook puts the same year, for the same country, at 1.088:
GET /api/v1/imf/revisions/weo
/DEU/NGDP_RPCH
?vintage=WEO_2025_OCT_VINTAGE
year latest Oct 2025 delta
2027 1.088 1.473 -0.385
2028 1.051 1.203 -0.152
2029 0.891 0.980 -0.089Source: IMF, World Economic Outlook, the latest edition against the October 2025 vintage, real GDP growth in percent, pulled 1 November 2026.
Nobody was wrong: a projection gets revised as the data under it changes. But a model trained before the revision will quote 1.47, cite the IMF, and be telling the truth about a number that is no longer the number. The same call for China returns -0.178 on 2027, and for the United States +0.238, so the revisions do not even point the same way. A stale price is obviously stale. A stale forecast looks exactly like a fresh one. If you cite a projection, cite its vintage: the vintage is its date.
Every economic figure has a name on it: the agency that issued it, the definition it uses, and the date it was taken. A model that quotes a figure without those three things is quoting something it cannot check, and the sentence reads just as confidently as one that is correct.
That is the job of Sugra Macro, one of the seven directions of Sugra Data. It returns the published record of whole economies, with its source and its date attached. The video at the top of this post walks through the record area by area.
Five areas, then a sixth that Sugra composes from them. For each area there is one real request below and the answer it returned, cut down to the lines that matter, with its source and its date. Everything was pulled from the live API on 1 November 2026. The two big examples from the video, the US economy and US trade with China, come after the areas.
Central banks and monetary
The Federal Reserve publishes its own balance sheet every week, broken down by what it holds. The composition matters more than the total: a balance sheet heavy in bills behaves differently as it runs off than one heavy in mortgage-backed securities.
GET /api/v1/fed/soma/history?limit=2
as of 2026-10-28
total 6,355,840,126,861
notes and bonds 3,600,742,227,800
mortgage-backed 1,876,203,782,906
bills 571,527,926,700
as of 2026-10-21
total 6,363,842,057,526
mortgage-backed 1,887,796,043,930
bills 567,874,926,700Source: Federal Reserve, the System Open Market Account, weekly, data as of 28 October 2026, pulled 1 November 2026.
In that week the total fell by about 8 billion dollars. Mortgage-backed holdings fell by about 11.6 billion, while bills rose by about 3.7 billion. The mix changed more than the small fall in the total suggests.
Central banks are not only the Federal Reserve. Sugra Macro carries the European Central Bank, and the central banks of England, Japan, Canada, Switzerland, Sweden, Norway, Poland, Czechia, Croatia, Australia, Malaysia, Brazil and Peru: thirteen besides the Federal Reserve and the ECB. It also carries about 800,000 more series from FRED, the Federal Reserve Bank of St. Louis database.
Statistical agencies
Statistics offices publish their own definitions, and the definitions differ from country to country. Sugra Macro carries the offices of 17 countries and territories, along with Eurostat, for inflation, GDP, unemployment, industrial output and retail sales.
Spain’s National Statistics Institute, the INE, publishes its consumer price index with the annual change for each month. It marks the latest month as a preview until the figure is final, and the answer carries that status with each value:
GET /api/v1/ine/cpi?nult=3
serie IPC290750, national, annual %
2026 M08 4.3 final
2026 M09 4.9 final
2026 M10 4.9 previewSource: Instituto Nacional de Estadística (INE), Spain, data through October 2026 (preview), pulled 1 November 2026.
The October figure is a preview, and the answer says so. A model that quotes it has the status it needs, not only the number.
Global economy and trade
Trade does not stop at the border, and neither does what moves through it. The IMF, the World Bank, the OECD, the WTO and UN Comtrade put their figures on a common footing. Routes are a different kind of record. Sugra Macro counts daily transits through 28 straits and canals, and it also carries the operators’ own logs, such as the Panama Canal Authority’s record of the transits it authorised.
GET /api/v1/maritime/chokepoints
/panama/transits
chokepoint Panama Canal
observed 2026-10-30 (report 10-31)
transits 32 in total
neopanamax 10 6 north, 4 south
super 16 9 north, 7 south
regular 6 3 north, 3 south
panamax plus 0
basis operator's own transit recordSource: Autoridad del Canal de Panamá (Panama Canal Authority), report of 31 October 2026 covering 30 October, pulled 1 November 2026.
The three classes add up to the total of 32, and the basis line is the point: this is the canal’s administrative record of transits it authorised, not a count inferred from vessel-tracking data.
Public finance and government
Who owes what, and to whom. The US Treasury publishes total public debt outstanding on every business day:
GET /api/v1/treasury/debt
record date 2026-10-29
total public debt outstanding
40,593,637,262,883
held by the public
32,712,190,190,967
intragovernmental holdings
7,881,447,071,916Source: US Treasury, Fiscal Data, record of 29 October 2026, pulled 1 November 2026.
Forty point six trillion dollars, and the split matters. About 32.7 trillion is held by the public, and about 7.9 trillion is held inside government accounts. The same area carries federal spending from USAspending, bills from Congress.gov, and TIC data on who holds US securities.
Fixed income
The bond market prices expectations, and the Treasury publishes its par yield curve every day. The curve for 30 October:
GET /api/v2/fixed-income
/treasury/yield-curve
curve: nominal par yields, 2026-10-30
1M 4.06 1Y 4.36 5Y 4.71
3M 4.16 2Y 4.56 10Y 4.94
6M 4.25 3Y 4.65 30Y 5.31Source: US Treasury, Daily Treasury Par Yield Curve Rates, 30 October 2026, pulled 1 November 2026.
The Federal Reserve’s target range is 3.75 to 4.00 percent, from FRED, as of 1 November 2026. Every tenor the API returns sits above the top of that range, and the lowest, one month, is 4.06. The curve climbs from there, to 4.94 at ten years and 5.31 at thirty.
Composed by Sugra
The sixth area is the one Sugra assembles. A country profile comes back in one request, built from the published sources above. Germany:
GET /api/v1/macro/country-profile
?country=DE
country Germany
population 83577136
gdp_yoy 0.012457
cpi_yoy 0.03263
core_yoy 0.02539
jobless_rate 0.041
yield_10y 0.03473
current_account_gdp 0.05498
gdp_usd null
policy_rate null
govt_debt_gdp nullSource: composed by Sugra (sugra-macro) from the published sources above, response dated 1 November 2026. Values are shown as the API returns them: rates and ratios come back as fractions, so cpi_yoy 0.03263 is 3.3 percent and yield_10y 0.03473 is 3.47 percent.
Three fields came back null: GDP in dollars, the policy rate and government debt. The profile shows the gap rather than filling it. The same area answers the reverse question too, one indicator across many countries in a single table.
The two big examples
The video closes with two questions. The first is how the US economy is doing. Three figures, each from its own source:
GET /api/v1/fred/series/UNRATE
?limit=2&sort_order=desc
Unemployment Rate, seasonally adjusted
2026-09 4.2 %
2026-08 4.1 %Source: Bureau of Labor Statistics, through FRED. Latest month September 2026, last updated 2 October 2026.
GET /api/v1/worldbank/bls/cpi-all
?years=2
CPI, all urban consumers
2026-09 336.94 12-month change 3.7 %
2026-08 334.98 12-month change 3.4 %Source: Bureau of Labor Statistics, consumer prices, September 2026, pulled 1 November 2026.
GET /api/v1/fred/series/DFEDTARL
?limit=1&sort_order=desc
GET /api/v1/fred/series/DFEDTARU
?limit=1&sort_order=desc
federal funds target range, 1 Nov 2026
lower 3.75 %
upper 4.00 %Source: Federal Reserve, through FRED (DFEDTARL and DFEDTARU), pulled 1 November 2026.
Unemployment at 4.2 percent, inflation at 3.7 percent on the twelve-month change, and a policy range of 3.75 to 4.00 percent. Each is a published figure with its agency and its date, and none of them is a forecast.
The second question is trade between the US and China. The Census Bureau’s balance with China, monthly, from March to August 2026. Partner code 5700 is China in the Census country list:
GET /api/v1/census/trade/balance
?partner=5700
&from=2026-03&to=2026-08
partner CHINA balance, USD
2026-03 -9,756,052,905
2026-04 -10,394,079,978
2026-05 -14,390,443,790
2026-06 -15,578,773,575
2026-07 -17,360,505,574
2026-08 -18,403,110,802Source: US Census Bureau, monthly, latest month August 2026, pulled 1 November 2026. The balance is exports minus general imports, as the response states.
The deficit with China widened in each of the five months from April to August, from about 9.8 billion dollars in March to about 18.4 billion in August.
Where this stops
We return the published record. We do not forecast, and where we return someone else’s forecast we label whose it is and which vintage.
Revisions mean today’s value for a past period may change tomorrow, so store what you saw and when you saw it rather than assuming a series is immutable. Frequencies differ across this direction, from daily yields to the weekly balance sheet and monthly prices, and joining them without stating the mismatch produces charts that imply a precision nobody has. National definitions differ even when field names match. And a preview is not a final figure: the INE marks its October value as one, and the answer carries that status for you.
Not investment advice, and not a substitute for reading the publisher’s methodology when the number is load-bearing. The full catalogue and its sources are in the Sugra Macro documentation, and every source named here is Tier S, I or A, listed in the source register.
Running it
H="x-api-key: $SUGRA_API_KEY"
B=https://sugra.ai
# the record, from the Federal Reserve
curl -sS -H "$H" \
"$B/api/v1/fed/soma/history?limit=2"
# from the statistics offices
curl -sS -H "$H" \
"$B/api/v1/ine/cpi?nult=3"
# public finance and the bond market
curl -sS -H "$H" \
"$B/api/v1/treasury/debt"
U=$B/api/v2/fixed-income/treasury
curl -sS -H "$H" \
"$U/yield-curve"
# one country, composed
C=$B/api/v1/macro
curl -sS -H "$H" \
"$C/country-profile?country=DE"Store the source and the date next to the value. Six months from now, the difference between “the Fed held 6.36 trillion” and “the Fed held 6.36 trillion on 28 October, per the SOMA record” is the difference between a guess and a citation.
Published official record. Not investment advice. Values pulled live 2026-11-01.