# How to research stocks: a step-by-step tutorial
> How to research stocks before buying, in six checks: business, numbers, filings, who is talking and their record, bull against bear, and timing. Dell walked through end to end.
By Maya Koeva · 2026-09-11
You keep seeing the same ticker. A friend mentioned it, it is all over your feed, it is up
20% this month. Before "should I buy it" comes a smaller question: how do you research a
stock in an order that does not eat a whole evening and still leave you guessing?
This tutorial is that order: six checks, cheapest first, each one allowed to kill the idea.
Then one real stock, Dell, walked through all six with the numbers we tracked this month.
If you do not have a name in mind yet, this is the wrong page. Start with
**[how to pick stocks](/learn/how-to-pick-stocks)** for a repeatable way to build a
shortlist, or **[how to know what stock to invest in](/learn/how-to-know-what-stock-to-invest-in)**
if you are choosing your very first one. This page assumes you already have a ticker and
want to know whether it survives scrutiny.
## Researching a stock is six checks in a fixed order
Most stock research guides hand you a pile of things to look at and no sequence. The
sequence is the useful part. Cheap checks go first, because they reject most ideas in
minutes and save the expensive reading for the few that pass. And one check that almost
every guide skips goes in the middle: who is talking about the stock, and whether those
people have been right before.
1. Explain the business in one paragraph.
2. Check the numbers that matter.
3. Read what the company says about itself.
4. Find out who is talking about it, and their record.
5. Read the bull case and the bear case side by side.
6. Check the timing.
Any check can end the research. Rejecting a name in ten minutes is the sequence doing
its job.
## Step 1: Explain the business in one paragraph
What does the company sell, to whom, and how does it get paid? What would have to be true
in three years for it to be worth more than it is today? Write the answer as one paragraph
in your own words, without the company's marketing copy.
If you cannot write the paragraph, you are guessing about a ticker. Stop here and read
until you can. If the paragraph comes easily, note what in it
is hard for a competitor to copy: a brand, a network, switching costs. That is the
company's [economic moat](/learn/what-is-an-economic-moat), and it is the first thing a
long-term holder is buying.
## Step 2: Check the numbers that matter
Four numbers, and the direction of each over the last eight quarters matters more than any
single reading:
- **Revenue growth.** Is the business getting bigger, and is the growth speeding up or
slowing down?
- **Margins.** Is the company keeping more of each dollar it earns, or less? Expanding
margins are the quiet sign of pricing power.
- **Debt against cash.** Who survives a bad year is decided here. A company funding
itself from its own cash flow can afford mistakes.
- **Valuation.** What are you paying per dollar of earnings? The
[P/E ratio](/learn/what-is-a-pe-ratio) is the fastest first look; for a fast grower,
check the forward P/E too.
All four live in the company's quarterly report and on any free finance site. This check
takes ten minutes and it is where most "hot" names fail: great story, shrinking margins.
## Step 3: Read what the company says about itself
Three documents, in this order:
1. **The latest earnings release and call.** Skip the headline beat. Stocks react to
[forward guidance](/learn/what-is-forward-guidance), which is management's own
forecast, and a company can [beat earnings](/learn/what-is-an-earnings-beat) and still
fall on a weak guide.
2. **The risk factors section of the annual report.** This is the bear case management is
legally required to admit. Read it before you read anyone's bull case.
3. **Insider transactions.** Executives sell for many reasons and buy for one.
[Insider buying](/learn/what-is-insider-buying) is not a signal on its own, but a
cluster of it after a drop is worth noticing.
## Step 4: Find out who is talking about it, and whether they have been right
Most guides leave this check out. It is the one that separates researching a stock from
absorbing the [mood around it](/learn/what-is-market-sentiment). Three questions:
**How much attention is there, and is it rising?** A mention count that doubles in a
week means something is happening. It does not mean the crowd is right. Treat rising
attention as a reason to do steps 1 to 3 carefully, never as a reason to skip them;
**[volume vs. signal](/learn/volume-vs-signal)** covers why the loudest name is rarely the best bet. And
read a low count for what it is: few of the accounts you follow are covering the name,
which is a gap in coverage, not proof that nobody cares.
**Where is the conversation happening?** [Reddit and X produce different signals](/learn/reddit-vs-x-stock-signals).
A name that is loud on one and silent on the other is a narrower story than one that is
loud on both.
**Who is doing the talking, and what is their record?** This is the question that matters,
and the one that is hardest to answer by hand. A thousand accounts repeating one thread is
one opinion with an echo. Find the loudest voices and check whether their past calls held
up before you weigh their current one. **[How to verify a track record](/learn/how-to-verify-a-track-record)**
covers the manual method.
This is the part of stock research Quantral was built for. It follows a curated set of
finance voices across Reddit and X, [grades every directional call](/learn/how-a-track-record-is-graded)
they make against what the price did next, and shows the record next to the post. The
[credibility score](/learn/what-is-a-credibility-score) that comes out of that grading is
what weights the conversation, so a name's [signal score](/learn/what-is-a-stock-signal)
reflects who is talking, not just how many.
One trap to name: concentration. If one account produces most of a stock's mentions, you
are reading one person, however loud. Check the spread before you trust the count.
**[How to use social signals without getting played](/learn/how-to-use-social-signals-without-getting-played)**
covers the rest of the traps.
## Step 5: Read the bull case and the bear case side by side
Write each in three lines. The bull case is usually easy to find, because it is what
brought you here. The bear case takes work, and finding it is the point:
**[bull case vs bear case](/learn/bull-case-vs-bear-case)** shows how to build both
fairly.
Then weigh them by step 4. A 7-to-1 bullish split among accounts with long graded
records is a different fact from 7-to-1 among accounts with no record at all. The
[sentiment breakdown](/learn/how-to-read-a-sentiment-breakdown) tells you the split; the
records tell you how much to trust it. If you cannot find a bear case anywhere, worry.
## Step 6: Check the timing
Three questions close the research:
- **What is ahead?** Earnings, a product launch, a regulatory decision: the
[catalysts](/learn/what-is-a-catalyst) that could move the stock in the next quarter.
- **What is already [priced in](/learn/what-does-priced-in-mean)?** Good news everyone
expects moves nothing.
- **Did the crowd arrive before the move, or after it?** Line up the mention spike
against the price. If attention landed after the jump, you are late to the story. You
may not be late to the stock, but you should know which one you are buying.
## How to analyze a stock before buying: Dell, end to end
Dell sits near the top of our 7-day list this week, scoring 87 (Strong) as of Thursday
night. It is a good test because the story is fresh, the conversation is split, and the
timing question has a clear answer. Every number below comes from the accounts we track,
and the cards are Dell's page in the Quantral app, so you can see where each check lives
(**[the app tutorial](/learn/how-to-use-quantral)** walks those screens in order).
**Step 1, the business.** Dell sells PCs and servers to businesses. The part that matters
now is AI servers: racks of GPUs assembled and shipped to data centers. The three-year
question is whether that order book keeps growing and whether Dell keeps any margin on it.
Easy paragraph. Pass. In the Quantral app, the About tab on Dell's page gives you the raw
material for it in one screen.
**Step 2, the numbers.** The figures the room quoted from the September 1 release: a
record $60.9 billion in AI server orders in the quarter, a $95 billion backlog, and
full-year revenue guidance raised to $192 billion from a prior range of $165 to $169
billion. Growth, check. The margin on those servers is the number to watch, and the bears
below know it. The Stats tab on the same page in the app covers the price side of this
check, the 52-week range and the returns, which tells you what you are paying relative to
the year. Revenue, margins and debt still come from the filings.
**Step 3, the filings.** Dell reported on September 1 after the close. The beat was large,
but the guidance raise was the story, which is the pattern step 3 predicts. The release
and the annual report are one tap away: the SEC filings link on Dell's About tab in the
app opens the company's own documents.
**Step 4, who is talking.** Over the 30 days to September 10, the accounts we track
produced 178 mentions of Dell, split 93 Reddit to 85 X, from 98 different accounts. The
loudest account was 15% of the total, so there is no single-voice echo here. Among the 27
accounts with at least ten graded calls, the split was 68 bullish to 9 bearish. Two of
those graded accounts, as their posts appear in the app's feed, the evening before the
print and the evening of it:
The loudest of those graded accounts:
| Account | Graded calls | Bullish on Dell | Bearish on Dell |
|---|---|---|---|
| @epictrades1 | 121 | 22 | 2 |
| @Mr_Derivatives | 210 | 14 | 1 |
| @pequityresearch | 67 | 8 | 0 |
| @RealSimpleAriel | 143 | 5 | 1 |
| @ThematicTrader | 257 | 3 | 0 |
*[Chart: Tracked Dell mentions per day (ET), August 12 to September 10: positive, negative, chatter, and noise. The wall is September 1, the evening of the earnings release.]*
**Step 5, bull against bear.** Across all 178 mentions the split was 98 bullish to 33
bearish, with the rest neutral. The bull case, in three lines: record orders, a $95
billion backlog against $192 billion of guided revenue, and a server refresh cycle still
ahead. The bear case, also from the room, and worth more than the bulls' volume suggests:
one graded account passed along a well-known TV analyst's view that Dell "will eventually
sell off despite a huge beat and guide"; a Reddit thread argued that a server company
guiding to fiscal 2027 is forecasting further out than anyone can see; and the
pre-earnings posts noted an implied move of about 10% already priced into the options.
Three real arguments. Pass, with the bear case written down.
Weighing those two sides by who said them is the analysis [Quantral's score](/learn/what-is-a-stock-signal) does for you. Dell's
page in the app shows it as two numbers: the week at 87 (Strong), because the graded
accounts were bullish and stayed that way after the print, and the day at 50 (Weak),
because Thursday's conversation was thin. The one-line reason under the rings says
which input is doing the work.
**Step 6, timing.** This is the check that complicates Dell. The stock fell 6.8% on September 1
to $425.00 on a weak tape, before the print. The mention wall hit at 4pm ET that day, the
hour the release landed: 23 bullish posts to 1 bearish in that single hour. The next
session Dell closed at $492.20, up 15.8%, and ran to $535.25 by September 9 before giving
back 5.4% on September 10.
*[Chart: Dell daily closes, August 12 to September 10. The low is September 1, the day of the release; the run that follows is the market repricing the guidance.]*
The other half of step 6 is what is ahead. The Events card on Dell's page in the app
lists what is scheduled, and right now that is one item:
The next earnings date is not on the calendar yet. When a company confirms one, it shows
up here, and that date becomes the next point where the crowd and the price get to
disagree again.
So the crowd reacted with the print, not before it. Anyone researching Dell this week is
researching a stock after its move, and the 87 score says credible attention is still on
the name, not that the next move is up. Dell passes steps 1 to 5. Step 6 says the
catalyst is behind you, which turns the decision into a different question: do you
believe the three-year paragraph from step 1 enough to hold through the next 10% swing?
Only you can answer that, and it is the right question.
## How the research changes for day trading, and for beginners
**Researching stocks for day trading** inverts the order. Steps 1 to 3 barely matter by
lunchtime; liquidity, volatility, and the state of the conversation right now are the
whole job. Our data is built for research at longer horizons and is not an intraday
trigger, so use it to pick which names to watch, not when to click.
**Researching stocks as a beginner** means running steps 1 to 3 in full before touching
4 to 6. Sentiment is the most seductive input on this page and the easiest to misread. If
step 1 fails, stop; there is no crowd loud enough to replace understanding what you own.
## The six-step checklist
1. **One paragraph on the business.** Cannot write it? Not a candidate.
2. **Four numbers, eight quarters.** Revenue, margins, debt, valuation. Direction over
level.
3. **Release, risk factors, insiders.** Guidance over the beat.
4. **Who is talking, where, and with what record.** Check the spread before the count.
5. **Bull and bear in three lines each.** Weighted by who said them.
6. **Catalysts, what is priced in, and whether the crowd is early or late.**
Run the six on every name that reaches you and let them reject most of what enters. The
stocks that pass a checklist you wrote while calm are the ones you can still explain when
the price is down and your feed has moved on.
---
*Quantral surfaces signals and context from public sources to support your own research.
Nothing here is financial advice or a recommendation to buy or sell. Past signals are
not indicative of future results.*