The Premium Scanner
Every way to narrow the market sits in one filter rail: a preset to start from, the filter sections you build with, and MAD Setups. A scan produces a shortlist, never a conclusion.
MadStockAlerts Research · Updated October 18, 2026
What to Take Away
- A preset picks the universe, the filter sections narrow it, and MAD Setups adds detection; all three sit in one rail and combine.
- Saved scans, the column picker and CSV export are on every member plan.
- MAD Setups needs Premium Stock Alerts.
- Every scan needs a liquidity constraint or the results are untradeable.
- Over-filtering feels like precision and is usually curve fitting.
MAD Academy Training Video · 0:54
A Filter, Not an Answer
The Scanner narrows thousands of securities to a readable list, and everything after that is research you still have to do.
This lesson is part of a MAD Academy plan.
The filter rail
Everything that decides which stocks you see lives in one filter rail, opened one section at a time from a row of section buttons that never moves. On a desktop the rail sits beside the results, or across the top of the page if you move it there; on a phone it is a panel that folds to a single line over the results. The line at the top of the rail states the scan: the preset it starts from, how many filters are on, and how many stocks match.
| Section | What it does | What it needs |
|---|---|---|
| Presets | Maintained screens for common questions. The scan starts from one, and every filter narrows inside it | Available to members |
| Profile, Fundamental, Technical, Extended Hours | Build your own query across fundamentals, technicals and price action | Tools for saved scans and export |
| MAD Setups | Nightly setup detection across the whole market | Premium Stock Alerts |
The scanner page serves more than one tier and adjusts what it offers rather than blocking the surface. A session with Alerts but not Tools reaches MAD Setups and not saved scans; the reverse holds for a Tools session.
What the filters reach
The filter sections are grouped by what they describe, and the grouping also says how current each number is. A live price and a figure from last quarter's filing can sit in the same screen, and knowing which is which is part of reading the result.
| Section | What it screens on | How current |
|---|---|---|
| Profile | Exchange, index membership (S&P 500, Nasdaq 100, Dow 30), asset type, sector, industry, market cap, price, volume, dollar volume, relative volume, shares outstanding and float | Price and volume are live; the rest changes rarely |
| Fundamental | Valuation ratios, dividend yield, EPS, return on equity and assets, balance-sheet ratios, short float and days to cover | Once a day, from reported financials |
| Fundamental, 13F | Held by Tracked Managers %: the share of outstanding shares held by the 107 institutional managers MadStockAlerts tracks. It is Institutional desk data, so the MAD Academy plan sees it locked | The last complete quarter; 13F filings are due 45 days after it ends |
| Fundamental, filed quarter | Revenue growth, EPS growth, and gross, operating and net margin, each against the same quarter a year earlier | The newest quarter filed with the SEC, so four times a year |
| Technical | Change, gap, RSI, MACD, moving averages, price ranges from 20 days to all history, performance over six windows from a week to a year, volatility, ATR, beta and VWAP | A daily sweep of closes; change and gap are live |
| Extended Hours | Pre-market and after-hours change, against the reference each session has: yesterday's close before the bell, today's close after it | Live during each session; off outside it |
A field is offered only when it can be measured across the whole market, small caps included. Earnings dates and ownership percentages are not screenable, because the data behind them is either a projection rather than a confirmed figure or covers only part of the market, and a filter over part of the market returns a list shorter than the truth.
Held by Tracked Managers % is not total institutional ownership. It counts only the 107 managers MadStockAlerts tracks through 13F filings, so it understates what every institution together holds, most of all in small companies. A blank means none of those managers reported the stock.
EPS growth is measured only when the year-ago quarter was profitable. A move from a loss to a profit has no meaningful percentage, so those companies are left blank rather than shown with a number that reads as a collapse or an explosion.
Building a scan that returns something usable
- 1Start with the universeMarket capitalization and average volume floors first. This is the constraint that decides whether results are tradeable at all.
- 2Add the condition you actually care aboutOne or two criteria that express the idea, not eight that express nothing.
- 3Read the countFour hundred results means the filter did not filter. Two means it was fitted to a foregone conclusion.
- 4Sort, then readA scan produces a list to examine, not a list to act on.
A scan with no liquidity floor returns thinly traded securities, because extreme readings are far more common where the price series is formed by very few trades. That is a property of thin data rather than a discovery.
Over-filtering
Adding criteria until the result set is small feels like precision and is usually curve fitting. Each additional condition is a further assumption, and a scan that returns two names has typically been shaped until it returned the names expected.
A useful discipline is to write the idea in a sentence first and then translate it into the fewest filters that express it. If the sentence needs eight clauses it is probably several ideas rather than one.
Scroll the chart sideways to see all of it.
What a scan is and is not
A scan narrows thousands of securities to a readable shortlist. Everything after that, reading the company, the filings, the liquidity and the structure, is work the scan has not done.
The scanner surfaces candidates; it does not evaluate them. Treating the output as an answer rather than a question is the most common way the surface is misused.
Reading the count as a diagnostic
Before any result is examined, the number of results says something about the scan itself. Treating that number as a diagnostic rather than as an outcome catches most badly specified scans immediately.
| Count | What it usually means | What to do |
|---|---|---|
| Several hundred | The conditions describe an ordinary state of affairs | Tighten the condition that expresses the idea, not the universe |
| 20 to 60 | A workable shortlist | Read it |
| Under 5 | Over-filtered, or a condition that is rarely satisfied | Remove the most recently added criterion and look again |
| Zero | Either genuinely nothing qualifies, or a criterion is malformed | Relax one condition at a time to find which one binds |
The fourth row is worth taking seriously rather than fixing. A scan that returns nothing has answered the question it was asked, and no securities currently satisfying a set of conditions is a legitimate finding about the market.
Loosening conditions until something appears is the most common way a scan is turned into a foregone conclusion. The conditions were the idea, and changing them until the answer is non-empty replaces the idea with the wish for a result.
What happens after the list
A scan is a hypothesis generator. It produces a set of securities that satisfy conditions, and every one of them still has to be examined for the things the conditions could not see.
- 1Check the calendarA scheduled event inside the expected window changes the risk of anything held through it, and no scan condition covers it.
- 2Look at the chartConditions describe values; the chart shows structure, and two securities with identical readings can look nothing alike.
- 3Check liquidity properlyAverage volume is a rough proxy. What matters is whether a position could be exited on a bad day.
- 4Find out whyA security satisfying a set of conditions did so for a reason, and the reason is in the filings or the news rather than in the scan.
- 5Discard most of itA shortlist that survives this intact was not a shortlist, it was a conclusion reached before the examination.
The step most often skipped is the last one. A scan that produces thirty names and a process that acts on thirty names has used the scanner as a decision procedure rather than as a filter.
Finished this lesson?
Progress tracking, subject trophies and the training videos are part of every paid plan.