No more scraping and hand-built spreadsheets: 17 years of ROIC in a single call — with the math laid open for you to audit.
455 publicly listed companies, 2010–today, straight from the public data of Brazil's securities regulator (CVM) and the Brazilian stock exchange (B3). Every indicator ships with its formula published, line by line. Indicators span the whole series; multiples only where price and a public share count existed — the year-by-year coverage is open. Every update clears 274 tests before it goes live. No black box, no return promises — just the number, and the path that produced it. Beyond stocks, the same rigour covers real-estate funds and macro series.
Numbers on this page are live. Data version 2026-09-01-supressao-declarada · page generated 2026-09-06 02:30 UTC. If this does not match /saude, you are reading a cached copy.
WEG's ROIC, year by year
Real data, straight from the API, right now — WEG is open for you to sample.
No token, no sign-up. Run it right here.
WEGE3 is 100% open on every endpoint. Hit the button and see the raw API response — the very same one your code will receive.
Python, JavaScript or the terminal — your call
WEGE3 needs no key. For the other 455 companies, just send yours in the X-API-Key header.
Four things none of the five APIs we compared expose
We compared five of them, with public sources and the date of the check — the comparison is open, including where they beat us.
The formula behind every number, line by line
Every indicator has a page explaining the calculation step by step — ROIC = NOPAT / invested capital, with each CVM account identified by its code. None of the five providers we compared publishes it at this level; here you can check it, and if you disagree, you know exactly where.
If one test fails, the update never ships
Before any release, 274 automatic checks run, and 155 of them ask whether the NUMBER ties out: the balance sheet balances, the income statement ties out, an extreme number gets flagged. The rest test the product. No test proves a number is right; these prove it ties out with the rest of the balance sheet — and flag where to trust it less. See the tests and a reconstructed example →
P/E at the price you could actually have paid
P/E, P/B and EV/EBITDA computed with the price of the first trading session from the day the report was actually published — never the Dec 31 price, because the statements only come out in Feb/Mar. Without it, the backtest buys at a price nobody could have paid that day — and the result rests on information the investor did not yet have.
Nothing is corrected in silence
When a company restates a report with different numbers, the record stays: old version, new version, date. You know what was known at each moment — and what changed afterward.
Three steps, zero meetings
Test without a key
WEGE3 is open on every endpoint — no sign-up, no card, no talking to anyone. Want all 455 right away? Create your free key (200 requests/day, no card).
Subscribe and get your key instantly
Checkout via Stripe, $9.90/mo. The key appears on the screen right after payment — no "awaiting approval" email, no contract, no lock-in.
Ship to production with the same API
The code you wrote in step 1 goes to production without changing a line of logic: add the X-API-Key header and swap the ticker. Works in Python, JavaScript, Excel, Sheets — anything that speaks HTTP.
Sample, Free and Pro
The free plan gives you a real API key — it's not a trial, it doesn't expire, it asks for no card.
| Feature | Sample no sign-up |
Free key instantly |
Pro $9.90/mo |
|---|---|---|---|
| Full WEGE3, all endpoints | ✓ | ✓ | ✓ |
| Methodology pages (every formula) | ✓ | ✓ | ✓ |
| 455 publicly listed companies | — | ✓ | ✓ |
| Annual indicators — ROIC, ROE, margins, growth, net debt/EBITDA (17 years) | — | ✓ | ✓ |
| Quarterly accounts and trailing P/E | — | ✓ | ✓ |
| Point-in-time multiples (P/E, P/B, EV/EBITDA) | — | ✓ | ✓ |
| Daily prices since 2010 | — | ✓ | ✓ |
| Restatement history | — | ✓ | ✓ |
| Bulk export — the whole dataset in CSV, one request per set (indicators, facts, multiples, prices, dividends, restatements, corporate actions, REITs) | — | — | ✓ |
| API key | not needed | instant, no card | instant |
| Request limit | sample | 200/day | 10,000/day |
Two sites show the same ROIC. Both are right.
And the numbers differ — because “ROIC” is not a definition, it is a family of choices. Without knowing which ones were made, you are not comparing companies: you are comparing methodologies.
Average or ending capital?
Dividing the year's profit by year-end capital flatters whoever grew. We use the average of this year and last — and we say so.
Whose profit?
Consolidated profit includes the minority share; profit attributable to controlling shareholders does not. Net margin and ROE from the same statement move depending on the choice.
Which 2015 number?
If the company refiled in 2019, “2015 profit” has two versions. Using the newer one to describe 2015 tells the story with information nobody had.
Which price?
A Dec 31 P/E pairs a report published in March with a December price. No investor could have traded on that information that day.
None of these choices is wrong. What is wrong is not saying which one was made — because then the number cannot be checked, only accepted. Every one of them is declared, and that is why the formula, the source account and the filing date travel with each value.
And when we are the ones who got it wrong. On 2026-08-15 AZUL showed up with a R$ 1.67 quadrillion market cap — a hundred times the country's GDP — and went through the entire database without tripping anything. We pulled the sector index offline, traced the cause, wrote the tests that were missing and rebuilt it. The post-mortem is published, with what was wrong and for how long.
Need the whole dataset, not a single query?
Rebuilding this dataset through the API costs about 1,824 requests — nine days on the free plan. Pro hands it over as nine CSV files, one request each: indicators, facts carrying the CVM account they came from, point-in-time multiples, daily prices, dividends, restatements, corporate actions, real-estate funds and foreign flow. 1,155,846 rows, counted right now.
Every row carries the ticker and the CVM code, so the sets join to each other and to your own database. That is what turns this from a source you query into a research database you own: backtests, your own screener, a valuation model, a local copy you can version.
Subscribe — $9.90/mo See the sets and how big each one is →
Nothing was removed from the free plan to build this: every endpoint still answers, company by company. Pro adds taking the dataset with you.
One plan. No fine print.
Less than half the entry plan of competing APIs — and the only one with an auditable methodology among the five we compared.
- 455 companies, 2010–today
- 10,000 requests per day
- The whole dataset in CSV — nine sets, one request each
- All indicators and multiples
- Daily and quarterly prices
- API key instantly
7-day money-back guarantee — not for you, we refund it.
Self-service payment and cancellation via Stripe. No lock-in. Already a subscriber? Manage subscription.
Paste this into Claude, ChatGPT or your own agent
The API returns clean JSON at predictable URLs — AI agents navigate it on their own. This prompt starts analyzing real data right away:
Query https://YOUR_DOMAIN/empresas/WEGE3/indicadores and analyze WEG's ROIC evolution year by year. WEGE3 is open; for other companies, send the X-API-Key header with your subscription key. The formula for each indicator is at https://YOUR_DOMAIN/metodologia.