How to Understand Any Stock from Zero, in Five Steps

$NVDA is the most-owned AI stock on the planet. I'd bet most people holding it couldn't explain how it makes money.

In this market, that gap can get expensive quickly.

So here's a thought experiment. Say I woke up tomorrow with zero market knowledge and had to learn how to read a stock properly.

I'd go in this order every time. The business first. Then the numbers. Then what I'm paying. Then the chart. Then a set of rules to stop myself from doing something stupid.

I learnt the hard way why the order matters. My first year making prices on a rates desk, I sized a view as if it were an edge.

The view was roughly right in the end. The P&L along the way was painful.

I'll use Nvidia as the worked example throughout. The aim is to show the method. I make no call on Nvidia either way.

Bear in mind that you can apply the same process to nearly every listed stock. You might find it particularly helpful after the recent AI rout, in which the Nasdaq dropped more than 10% from its highs.

Prices are as at Monday's close on 2 November 2026. Consensus estimates are as at 3 November.

Nvidia is close to 8% of the S&P 500 on its own. The ten biggest names are about 39%.

If you own an index fund, you already own some Nvidia.

Its price also moved fast last week. On Friday 30 October, the Nasdaq fell more than 700 points.

Nvidia fell 4.2% to $204.64 that day. That left it 14.5% below its record close of $239.24 on 6 October.

On Monday it slipped another 1.3% to $202.06, 15.5% below the record. A move that big in a week tests whether you know the business well enough to judge it.

The question. What does the company sell? Who pays for it?

How to answer it. Read the first few pages of the annual report. Then read the segment table in the latest quarterly filing, which splits revenue by business line. If you still can't explain the business in two sentences, the later steps won't mean much.

Worked example: Nvidia. In the quarter to 26 July 2026, revenue was $96.2 billion. Data Center made up $89.0 billion of that, about 93%.

The biggest buyers are the Hyperscalers. Hyperscalers are the largest cloud providers, such as Microsoft, Amazon, Alphabet and Meta. They more than doubled their spending with Nvidia from a year earlier.

One customer accounted for 16% of revenue in the quarter.

Then check where the customers' money comes from. CapEx, or capital expenditure, is the money companies spend on data centres and equipment. Amazon, Alphabet, Microsoft and Meta expect combined CapEx of about $720 billion to $745 billion this year.

A large share becomes Nvidia revenue. More of it now comes from debt. Microsoft sold $25 billion of bonds on 29 October, with one part running to 2066 at 6.35% interest.

My two sentences on Nvidia: it sells AI chips plus the networking that connects thousands of them. Most of the money comes from a few very large cloud companies, which increasingly borrow to pay.

On any stock. Write down two figures: the biggest segment's share of revenue and the biggest customer's share. For Nvidia those are 93% and 16%. Filings usually disclose any customer above 10% of revenue, so search the filing for "10%".

The question. Is revenue growing? How much of each sale does the company keep? Does profit turn into cash? Could the balance sheet take a bad year?

How to answer it. Run four checks:

Worked example: Nvidia. Revenue rose 106% from a year earlier to $96.2 billion. Guidance, the company's own forecast, is about $108 billion for the quarter to October, give or take 2%.

Gross margin was 75.0%, so Nvidia keeps about 75 cents of every sales dollar before running costs. Management guides it lower from here because memory prices have risen sharply.

First-half operating cash flow was $74.4 billion against $4.4 billion of equipment spending. That leaves about $70 billion of free cash flow in six months.

Nvidia holds $56.6 billion in cash and short-term debt securities. Its debt is $33.5 billion of senior notes, a type of bond.

When the checks disagree. The four checks rarely all look good at once. When one looks weaker, find the reason and write it down as the stock's weak point.

A stock can keep rising for months while the market ignores its weak point. Once the market starts to focus on it, re-check your thesis, your reason for owning the stock. That re-check is a form of risk management.

Nvidia's weak point is the gross margin. Growth and cash are strong. The margin is slipping: 75.0% last quarter, 74% guided for the quarter to October, then 71% to 72% for the quarter to January.

Two business numbers would show the market starting to focus on it. The first is a gross margin guide below 71% at the results on 17 November.

The second is revenue below $105.8 billion for the quarter to October, the bottom of the guided range. That would put growth at 86%, down from 106% last quarter.

Some investors pair those with a price level from the chart. Step 4 shows one way to set it.

The accountant in me also watches receivables, the money customers owe for goods already delivered. The filing says they rose partly on "extended payment terms on large multi-quarter agreements". Nvidia also owns stakes in companies that buy its chips. Neither item changes the numbers today. If CapEx slowed, I would check both first.

On any stock. Write the weak point in one line, with the number that would trigger a re-check.

For Nvidia: "Weak point: gross margin, on memory costs. Trigger: a guide below 71% or revenue below $105.8 billion."

The question. How much growth does today's price already assume?

How to answer it. Every share price is two numbers multiplied together: earnings per share (EPS) and the multiple the market pays for them. Split the price into those two. Then work out the growth the price needs.

Worked example: Nvidia. Every multiple here is the share price divided by consensus EPS for this fiscal year. Nvidia's fiscal year (FY2027) ends in January 2027. Consensus EPS is $9.31 for FY2027, $15.91 for FY2028 and $21.38 for FY2029.

$202.06 divided by $9.31 is 21.7. Nvidia trades at 21.7 times this year's expected EPS.

The principle. If the multiple holds, the price grows with EPS. A rising multiple adds to the return. A falling one subtracts.

Consensus has EPS rising from $9.31 to $21.38 by FY2029, a 130% rise. I count that as two years because FY2027 ends in three months. It works out at 51.5% a year, the compound annual growth rate (CAGR).

At an unchanged 21.7 times, the price becomes $464, also 130% above $202.06.

Now cut the multiple to 13 times. Earnings rise 2.30 times while the multiple falls to 0.60 of its old level. The two multiply to 1.38, so the price rises 38% to $278.

Reverse the multiple. Pick an illustrative multiple for a mature business. I use the S&P 500's 18.7 times the next 12 months of earnings. It is market-cap weighted, so companies with the largest total share value carry the most weight.

$202.06 divided by 18.7 is $10.81. At today's price, Nvidia needs EPS of $10.81 by FY2029 to trade at 18.7 times.

From this year's $9.31, that is growth of 7.7% a year for two years. Consensus expects 51.5% a year.

The gap shows how much of a slowdown today's price already allows for. Steps 1 and 2 help you judge which figure is closer.

I remember Apple at about 10 times earnings ten years ago, when plenty of people called it cheap for a reason. A low multiple on fast growth usually means the market doubts the growth will last.

Move both numbers. Each implied price is the multiple times the EPS assumed. The multiples are illustrative.

The last row shows the risk to the EPS figure itself. Flat earnings at 13 times give $121.

On any stock. Divide the price by this year's consensus EPS to get the multiple. Divide it by a mature multiple, such as 18.7, to find the EPS the price needs. Compare that growth with consensus.

The question. How are buyers and sellers behaving right now? At which price would someone step aside to limit losses?

How to answer it. Start with the 50-day and 200-day moving averages, the average closing price over that many trading days. Then check the volume profile, which shows where the most shares changed hands.

Worked example: Nvidia. On Monday, Nvidia closed at $202.06, 1.2% below its 200-day average of $204.46. The 50-day average is higher, at $221.50, so the short-term trend has turned down. The 200-day average is still rising. It was $201.02 a month ago.

Over the past year, the heaviest trading was between $180 and $190. Far fewer shares changed hands between there and today's price.

One caveat on the 200-day line. We tested closes near it across the 126 names in our coverage.

Since 2000, 13,984 closes sat within 3% above the 200-day average. Five sessions later, the price was higher 54% of the time. Any random five-session window since 2000 was also higher 54% of the time.

On its own, the line says very little about next week. Levels help more as a way to limit losses in a downtrend. It helps to write that price down before the stock gets there.

The Nvidia levels below only illustrate the method. I make no call on where the stock goes next.

One way to set a step-aside rule. Start with two checks on the long-term trend. The weekly close should sit above a rising 40-week moving average, the weekly version of the 200-day. Each pullback on the weekly chart should also bottom above the last one. Each of those bottoms is a swing low.

Nvidia passes both checks by a small margin. Last week closed at $204.64, above the 40-week average of $204.15. The swing lows rise: $164.27 on 30 March, $189.80 on 29 June, then $190.01 on 29 July.

Setting the level. For example, someone using the 29 June low would step aside on a weekly close below $189.80. That sits 6.1% below Monday's close.

A weekly close filters out a fall that reverses within the day. On 30 October, Nvidia traded as low as $200.92 but closed at $204.64.

A tighter version. Someone wanting to step aside sooner could use a weekly close below the 40-week average instead. That is about $204.4 counting Monday's price.

Adding valuation. At $189.80, with FY2027 consensus EPS unchanged at $9.31, Nvidia would trade at 20.4 times this year's EPS. That compares with 21.7 times today. The market would be paying less for the same earnings.

On any stock. Find the last two swing lows on the weekly chart. If the later one is higher, the lower of the pair is one possible step-aside level.

The question. Which written rule decides when to hold the stock? How did that rule do on history after costs?

How to answer it. Write the rule down before you need it. Test it on as much history as you can, with trading costs included.

Worked example: Nvidia. Take the simplest trend rule I know. Hold Nvidia when it closes above its 200-day average. Move to cash when it closes below, trading at the next day's close.

I ran it from 2 January 2001 to 30 October 2026. Each move in or out costs 5 basis points (bp), with no interest on the cash. A basis point is 0.01%, so 5bp is 0.05% of the amount traded.

At 5bp a trade, the rule averaged 33.3% a year from 2001 to 2026.

Holding the stock throughout averaged 33.2% a year over the same period.

At 20bp a trade, the rule averaged 32.3% a year from 2001 to 2026.

Buy and hold had a worst fall of 90%. The rule's worst fall after costs was 55%.

The rule switched in or out 121 times. It sold 37 of its 61 buys again within 20 trading days. Every one of those 37 lost money. Their average loss over 2001 to 2026 was 3.8%.

One rule on one stock proves little. Still, a simple written rule matched buy and hold after costs while cutting the worst fall to 55%. A 90% fall is the kind that makes people sell at the bottom.

That's why we run four models behind our Multi-model Signals. When the story and the price disagree, I want a second opinion from outside my own gut.

On any stock. Test one rule and record four numbers: the yearly return after costs, the worst fall, the number of switches and the losses from quick reversals.

Each step produces a line or two. Together they make a one-page note to re-check on any date. Here is Nvidia's.

Save this. Each line has Nvidia's numbers as the worked example.

If you're starting from zero, our Academy on yxinsights.com explain each step in plain English. Our Multi-model Signals and Systematic Portfolio cover the systematic step.

DISCLAIMER: This newsletter is intended for educational purposes only. Any information or analysis in this note does not constitute an offer to sell or a solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice, nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.