GeminIQ Website Logo
Subscribe

Zacks Alternative: Filed Numbers, Not Estimates

Chad Hartman

By

Published · Last updated

Zacks has the cleanest thesis of any research firm on this list, and it has held it since 1978. Len Zacks concluded that earnings estimate revisions were the most powerful force acting on stock prices, and the Zacks Rank has expressed that single idea ever since — aggregating revisions across roughly 3,000 analysts and more than 200,000 estimates, scored on components including how many analysts are revising, by how much, and how results have compared to consensus.

Start your 7-day free trial →

Premium runs $249 a year and Ultimate $2,995, with research reports, screens, industry rankings, and model portfolios layered on top of the Rank.

Here is what the system contains, stated plainly: no as-filed data whatsoever. Every input to a Zacks Rank is a forecast, a change in a forecast, or a comparison against a forecast. It is a measurement of analyst behavior, and that is not a criticism — it is the design.

Table of Contents

A Revision Is Information About Analysts

Start with what an estimate revision actually is.

An analyst covering a company publishes an earnings forecast. Something changes their view — a data point, a channel check, a competitor's results, a conversation, a change in their own model's assumptions — and they publish a new number. The revision records that an analyst changed their mind and in which direction.

That is a real and tradeable phenomenon, and the case for it is well documented: revisions cluster, other analysts follow, and prices tend to move with the aggregate. Nothing here disputes that.

What a revision does not record is anything the company did. The business did not change when the analyst updated a spreadsheet. Reported results did not move. The revision is a fact about a forecast, and the forecast is a fact about a person.

The distinction gets blurred because the language of estimates borrows the language of results. An "earnings estimate" sounds adjacent to earnings. One is a number a company filed under officer certification. The other is a number a stranger typed into a model.

Beating Consensus Is Not a Statement About the Business

Earnings surprise deserves particular scrutiny, because it is the most misread figure in retail investing.

A surprise measures reported results against consensus expectation. It is a relative measure, and its denominator is a forecast that moves. That produces outcomes that look strange until the mechanics are clear.

A company can beat consensus while revenue declines, because the estimate had been revised down far enough. A company can miss consensus while growing strongly, because expectations ran ahead of it. Two companies with identical filed results can post opposite surprises purely because analysts approached them differently. And a company whose estimates were cut heading into a quarter has an easier bar than one whose estimates rose, which means the surprise partly measures how expectations were managed.

None of that makes surprise data useless — how a stock trades around a beat or a miss is a real market phenomenon, and the stock price reaction after earnings study covers what the filed record shows about those moves. It makes surprise a measure of expectation-setting rather than a measure of performance.

The filed statement is the absolute number. The surprise is the gap between that number and what someone predicted.

Coverage Determines Whether You Get a Signal at All

An estimates-based system inherits a hard structural limit: it requires analysts.

Zacks tracks thousands of analysts across a covered universe in the low thousands of companies. Outside that set the machinery has nothing to work with. A company with two analysts produces a fragile consensus that one revision can swing. A company with none produces no Rank at all.

Coverage correlates with size, liquidity, and institutional interest — which means the estimates layer is thinnest exactly where a fundamental investor is most likely to be looking. The small, unfollowed, and structurally ignored are the population where independent work has the best chance of finding something, and they are the population an estimate revision system cannot see.

Filings do not depend on anyone paying attention. Every registrant files on the same schedule whether or not a single analyst has ever written about it.

Estimates Change, Filings Do Not

There is a durability difference that matters for anyone building a record of their own reasoning.

A consensus estimate is a moving object. It is revised continuously, it is superseded by the next quarter's, and the version that existed when you made a decision is frequently not recoverable afterward. Reconstructing why a stock looked attractive on a given date is difficult when the central input has since been overwritten.

A filed 10-K is fixed. It was submitted on a date, under signatures, with an audit opinion attached, and it stays exactly as filed. When a company restates, the restatement is itself a filing — the original remains in EDGAR, and the difference between them is visible and dated.

For a process that has to be reviewed later, that permanence is worth more than it sounds. Analysis built on filings can be reconstructed. Analysis built on a consensus that has moved cannot be.

What the Filed Record Answers Instead

GeminIQ contains no estimates, no ratings, and no consensus. It extracts 10-K and 10-Q data directly from SEC EDGAR, preserves each company's own reported line item structure, and keeps the XBRL tag attached to every value.

Financial Statements show a company's own captions across quarters and years, so reported results can be read as a record rather than measured against a forecast. Visualizations chart the reported structure over time. Custom Tables assemble the specific reported items a question turns on. Calculated Metrics including Earnings Per Share Growth, Revenue Growth, and Net Profit Margin are computed from as-filed inputs, which means growth is measured against what the company previously reported rather than against what anyone expected.

The coverage universe is every SEC filer, because eligibility is a filing requirement rather than an analyst's decision.

Two Timeframes, Two Questions

The honest reconciliation is that these systems operate on different clocks.

Estimate revisions are a short-horizon phenomenon. They describe expectations shifting now, and the trading implications play out over weeks to a few quarters. That is a legitimate strategy with decades of research behind it, and an investor operating on that horizon is right to track revisions.

Filed fundamentals operate on a longer one. What a business earns on capital, whether margins are durable, how the balance sheet behaves under stress, and whether management allocates capital well are questions that resolve over years, and they cannot be read from a revision at all.

For the other side of the forecast layer — where estimates feed a fair value rather than a rank — the Morningstar alternatives and Finbox alternatives posts cover what a model inherits from consensus.

The mistake is running one clock while believing you are running the other. A Rank tells you which way expectations are moving — and only the filings tell you what the company actually reported, which is the number that is still true after everyone has moved on.

Frequently Asked Questions

What are the best Zacks alternatives?

It depends on what you are replacing. For earnings estimate revisions and consensus data, the substitutes are other estimates providers, all of which depend on analyst coverage. For reported fundamentals, the alternative is a platform that extracts from SEC EDGAR and preserves as-filed line items with XBRL tag traceability — which contains no estimates at all, by design.

What is the Zacks Rank based on?

It is built on analyst earnings estimate revisions, scored on components including how many analysts are revising estimates and in which direction, the magnitude of those revisions, and how reported results have compared to consensus. Every input is a forecast, a change in a forecast, or a comparison against one.

Are earnings estimates the same as reported earnings?

No. An estimate is an analyst's forecast, published before results and revised continuously. Reported earnings are filed with the SEC in a 10-K or 10-Q, prepared under GAAP, audited, and certified by company officers. An estimate is a prediction about a company; a filing is a statement by one.

Can a company beat estimates and still be shrinking?

Yes, and it happens routinely. A surprise measures results against consensus, so a company whose estimates were cut sharply enough can report declining revenue and still beat. The absolute performance is in the filing; the surprise measures the gap between that performance and what was predicted.

Start your 7-day free trial →

Wall Street's data. Main Street's price.

Institutional terminals charge thousands a year for as-filed accuracy. GeminIQ gives you the same thing for a fraction of the cost: financials built directly from raw SEC EDGAR filings, not third-party APIs, with full XBRL traceability back to the original 10-K or 10-Q. No normalized guesswork, just calculated metrics, charts, screeners, and watchlists built on numbers exactly as the company reported them. Start researching now at GeminIQ.com.

Data Used / Sources

  • Fundamental data sourced from XBRL-tagged SEC filings via GeminIQ.
  • Zacks Rank methodology, founding history, analyst and estimate counts, covered universe, and Premium and Ultimate pricing reviewed August 2, 2026 from Zacks published materials and third-party platform reviews. Verify current pricing before republication.

Disclaimer: The content in this blog is for educational and entertainment purposes only and does not constitute financial, legal, or tax advice. Investing involves risk, including the loss of principal. The views expressed are my own and not intended as financial advice or a guarantee of future performance.