A Compustat / WRDS Alternative for Point-in-Time Fundamentals, Honestly

If you searched "Compustat alternative," there is a good chance something specific happened: you graduated and your WRDS login stopped working, your institution does not subscribe, or you are trying to ship something small and cannot justify an institutional agreement to do it.

Let us be blunt before anything else. Compustat is the standard, and this page is not going to pretend otherwise. If you have WRDS access, use WRDS. Nothing here replaces it. What follows is an honest account of the narrow slice we do cover, and a clear statement of everything we do not.

We build Tradevo Data: point-in-time US equity fundamentals sourced from SEC EDGAR (public domain), delivered as one JSON endpoint at $29/mo.

What each product actually is

S&P Compustat, typically accessed through WRDS, is the institutional reference dataset for accounting data. Global coverage, history reaching back decades, quarterly and annual, thousands of standardized items, and — critically for backtesting — a genuine point-in-time product (Compustat Snapshot / the point-in-time database) that preserves what the data looked like on a historical date rather than only what it says now. It is the dataset published finance research is built on, and it is normalized across companies in ways that raw filings are not.

Access is institutional. Pricing is negotiated, not listed, and typically runs in the five-figure range annually for a university or firm subscription. Individual self-serve access is not really the model.

Tradevo Data is a deliberately narrow budget tier. We cover 5,170 US companies, 633,394 annual point-in-time rows, 16 annual concepts (Revenue, NetIncome, Assets, StockholdersEquity, OperatingCashFlow, EPSDiluted, DilutedShares, GrossProfit, OperatingIncome, PretaxIncome, IncomeTaxExpense, CapitalExpenditures, CashAndCashEquivalents, CurrentAssets, CurrentLiabilities, NetPPE), up to 12 fiscal years, plus 1,090,751 quarterly rows across 5,377 companies for seven quarterly concepts. Annual values come from 10-K/10-K/A filings; quarterly values come from 10-Q filings with supported Q4 values reported or derived and labelled. US only. Every figure comes straight from EDGAR with the date it became public attached.

That is roughly 0.5% of what Compustat gives you, and we would rather say so on our own comparison page than have you discover it after paying.

Side-by-side

See the lookahead leak on real filings — 40 companies, 6,969 rows, CC0, no signup. Three minutes in a notebook, before you decide anything.

Compustat and WRDS terms change and are negotiated per institution. Verify current specifics with S&P Global and WRDS directly — do not take our summary as authoritative about someone else's product.

When Compustat is the right answer

Most of the time, honestly. Use Compustat if:

• You are doing publishable research. Referees expect Compustat. A paper built on a 40-large-cap EDGAR extract will get desk-rejected, and rightly.
• You need TTM, a broader quarterly concept set, or institutional lineage. That remains Compustat territory; Tradevo Data quarterly v1 intentionally covers seven quarterly concepts with labelled Q4 derivations.
• You need pre-2013 history. Our 12 fiscal years reach back about a decade. Anything spanning 2008, the dot-com era, or multiple cycles needs real depth we do not have.
• You need inactive and delisted companies. Our coverage is active-company focused, which means a long-horizon backtest on our data carries survivorship bias. That is a real limitation, not a footnote.
• You need non-US companies. We are US-only.
• You need standardized fields across filers. Compustat's normalization is decades of accumulated work. We give you the raw XBRL tag and let you see exactly where each number came from, which is a different trade — better for provenance, worse for cross-company consistency.
• You already have WRDS access. Then this is not a decision. Use it.

If any of those describe you, go and use Compustat. We would rather lose the sale than have you build something on data that cannot support it.

When we actually fit

There is a real gap Compustat does not serve, and it is the one we are built for:

1. You lost institutional access. The single most common reason people search this term. You did your thesis on WRDS, you graduated, and now the same query costs more than your rent. If you need annual US fundamentals for a personal project, $29/mo self-serve is a different category of decision from an institutional agreement.

2. You need to ship something this week. WRDS access involves an institution, an agreement, and a login someone else controls. Our key is issued the moment Stripe clears. That difference matters when you are prototyping rather than researching.

3. You need to publish or cite the underlying numbers without a licensing conversation. Compustat is licensed data with genuine restrictions on redistribution. Our source filings are SEC EDGAR, which is public domain, and our free sample is CC0 — you can put it in a public repo, a paper, or a client deliverable. The paid API and dataset are licensed for your own internal and commercial analysis; bulk redistribution or resale of the dataset is not permitted (see our terms, §2). If your use case is republishing a whole dataset, the CC0 sample and the public methodology are the parts built for that.

4. You want to audit the data before trusting it. Every row carries first_filed (when the value became public), original_value (as first reported), latest_value (current revision), a restated flag, and a per-row qa_status — problems are flagged rather than silently dropped. We label 37,941 restatements across the dataset. The methodology and 6,969 sample rows are public on GitHub with no signup, so you can diff us against whatever you use now before paying anything.

5. You want to see the raw tag. We do not normalize. If Broadcom switched from NetIncomeLoss to ProfitLoss mid-history, you see that, flagged. Normalization is usually a feature; when you are debugging why a series stepped, it is the thing hiding the answer.

One thing worth knowing regardless of what you buy

Building this dataset surfaced a failure mode that is easy to miss and does not depend on which vendor you use.

In our 40-company sample, 189 values differ between what was first filed and what EDGAR reports today. Twenty of those are not accounting restatements at all — they are retroactive split adjustments. Amazon's FY2020 diluted share count reads 510,000,000 as originally filed and 10,198,000,000 in EDGAR today, because the 2022 twenty-for-one split is applied backwards through history. Both numbers are correct; they are denominated in different shares. The same pattern covers AAPL, NVDA, TSLA, WMT, NFLX, AVGO and NKE.

This is the case that defeats the defence everyone already knows. Waiting for the filing date does nothing, because the filing date is right — the number is simply denominated in shares that did not exist on it. Pair today's share count with a contemporaneous price and market cap is wrong by exactly the split factor.

Worth checking in whatever pipeline you already run, including a Compustat one.

Try it without paying

The free sample — 40 large caps, 6,969 point-in-time rows, full methodology, no signup, CC0 — is on GitHub. Diff it against your current source before trusting a single row of ours.

If it fits, the free API key arrives by email in a minute (250 requests/day, no card); the $29/mo plan adds bulk download, the universe snapshot and 5,000 requests/day — cancel anytime. If it does not fit — and for a lot of Compustat users it genuinely will not — the sample is still free and still yours.

Not investment advice. Compustat and WRDS pricing, coverage, and licensing are negotiated per institution and change over time; verify current terms with S&P Global and WRDS directly rather than relying on this page. Tradevo Data figures stated as of August 2026.