What should I research before selling cash-secured puts on a stock?

Research the business you might be assigned as well as the option. Check how the company earns its operating income segment by segment, how its margins build from gross to operating, whether the popular bull story survives the latest filings, what the bear case is (customer concentration, capex against free cash flow, debt, dilution), and what valuation the put strike implies. A put is a standing offer to buy the shares at the strike, so the question is whether you would be glad to own the company at that price.

Why isn't a wheel screener enough on its own?

A screener works from prices, volatility, open interest and financial ratios. From those you can check that a put pays at least what the stock actually moves, that the chain is liquid, and that no earnings land before expiry. Segment notes and earnings calls are outside its inputs. In the four examples here, every popular narrative had at least one part that inverted once the filings were read. None of those inversions would show up in a screen.

What is a margin walk and why does it matter to a put seller?

A margin walk traces revenue down to operating income: gross margin, then operating expenses, then operating margin, on both a GAAP and an adjusted basis. It shows which segment actually makes the money and what the adjusted figures leave out. For a put seller it answers whether earnings are growing because the business is getting better or because a cost has been moved out of the headline number, which matters if you end up owning the shares.

How do I judge a put strike on fundamentals rather than delta?

Convert the strike into a valuation. Divide it by guided or forward earnings to get the multiple you would pay on assignment, note how far it sits below the 52-week high, and ask what you would actually own afterwards, for example two companies after a spin-off, or a share count that is set to grow. Delta tells you the chance of assignment. The multiple tells you whether assignment is a price you would choose.

← Writing

Fundamentals

What to Research Before Wheeling a Stock: Four Worked Examples

27 September 2026 · 10 min read

A wheel screen shows whether a put is fairly priced. It does not show whether you want to own the shares at the strike. On the wheel, that second question matters more. Before selling a put, research how the company earns its operating income, how its margins are built, whether the popular story survives the filings, what the bear case is, and what the strike implies as a valuation.

This post works through that research on four names that passed a volatility-and-liquidity screen on 25 September 2026: TTM Technologies (TTMI), Flex (FLEX), Intel (INTC) and Nebius (NBIS). They are examples of the method, not a list of trades. Every figure is point-in-time, taken from company filings and releases, and will have moved by the time you read this.

Not investment advice. This is an example of our own research process on four real companies. It is not a recommendation to buy, sell, or sell options on any of them. Do your own research, check the latest filings yourself, and decide what suits your own situation.

Why a screen is not enough

The four names passed a screen on price data. Each put paid close to or above what the stock has actually been moving, each chain had open interest in the hundreds or more, and none had earnings before expiry. Those are the right filters for pricing. They say nothing about the business.

The filings told a different story. Each of the four had a popular narrative with at least one part that inverted. A defence story turned out to be a shrinking share of revenue. A diversified manufacturer's growth engine turned out to depend on two customers. A foundry's growth was mostly the company buying from itself. A gross margin excluded the largest cost. None of this appears in a screen.

The research, in order

What the business is. Revenue by segment or end market, for the latest fiscal year and the latest quarter.
How it earns operating income. Operating income by segment. Which part makes the money and which part loses it.
The margin walk. Gross margin, operating expenses, GAAP operating margin, over three years. Then the adjusted figure and what it excludes: stock compensation, amortisation, restructuring, depreciation.
The bull case. Two to four pillars, each with its mechanism in one sentence and a primary source you actually opened.
Where the story inverts. Check each pillar against the filing. A search summary is not a source.
The bear case. Customer concentration, capex against free cash flow, debt, dilution and any cheaply acquired stakes, China and tariff exposure.
Management's targets and the next earnings date, from the company's own investor relations page.
The strike as a valuation. The multiple of earnings the strike implies, the distance from the 52-week high, and what assignment would leave you owning.

TTM Technologies: circuit boards for AI servers and defence

TTM makes printed circuit boards. In Q2 2026, revenue was $1.0 billion, up 37% from a year earlier. Data centre and networking was 40% of sales, up from 29%. Aerospace and defence was 37%, down from 45%.

	FY23	FY24	FY25	Q2 2026
Gross margin	18.5%	19.5%	20.7%	21.1%
GAAP operating margin	1.9%	4.8%	9.1%	10.9%
Segment margin, defence / commercial	9.1% / 14.7%	12.4% / 14.1%	14.2% / 15.1%	16.7% / 18.1%

This is a manufacturer with about a 20% gross margin. The operating margin rose from 2% to 11% because AI boards carry more layers and higher prices, operating costs were spread over more revenue, and one-off charges rolled off. The commercial segment now earns a higher margin than defence.

The bull case. Data centre sales grew 91% in Q2. Defence work is long-cycle, with $1.7 billion of program backlog, and a new US plant for advanced boards is due to start volume production in the second half of 2026. Management guided a 13% to 15% adjusted operating margin for 2026.

Where it inverts. The popular framing is a defence-grade, US-made supplier. The filings show the growth and the higher margin coming from cyclical AI boards, largely made in China and Malaysia. China held about 40% of the company's long-lived assets at the end of FY25.

The bear case. The top ten customers were 55% of sales, up from 41% two years earlier. Free cash flow is about zero against $300 million or more of annual capex. Around $1.6 billion of new debt is funding three acquisitions, and both the US and China have tariffs on circuit boards.

The strike question. A put about 20% below the late-September price sits near 21 times guided 2026 adjusted earnings, on a stock already about 42% below its 52-week high.

Flex: contract manufacturing, with a power business being spun off

Flex had $27.9 billion of revenue in the year to March 2026, across three segments: general manufacturing ($11.1 billion), regulated manufacturing for medical and automotive ($10.2 billion), and cloud and power infrastructure ($6.6 billion, up 35% in the latest quarter).

	FY24	FY25	FY26	Q1 FY27
Gross margin	7.1%	8.4%	9.2%	9.4%
GAAP operating margin	3.2%	4.5%	4.9%	4.9%
Adjusted operating margin	–	5.7%	6.3%	6.7%
Segment margin, general / regulated / cloud	3.9 / 5.1 / 9.4%	4.8 / 5.2 / 10.2%	5.4 / 6.0 / 9.2%	5.2 / 6.6 / 9.7%

Each dollar of sales keeps five to seven cents of operating profit. Cloud and power earns the most, near 10%, but its margin fell last year, and most of the improvement came from the other two segments.

The bull case. Flex plans to spin cloud and power into a separate listed company, Axiom, targeted for the first quarter of 2027 and guided to grow 65% to 75% in the year to March 2027. The argument is that a fast-growing power business earns a higher valuation trading on its own. Flex reached its 2027 margin target a year early.

Where it inverts. The growth engine is not diversified. Two customers were 64% of the cloud and power segment's revenue last year, and the top ten were 87%. Most of the segment is server rack integration and cooling ($4.5 billion); power equipment is $2.1 billion.

The bear case. Capex is guided to roughly triple, to $1.5 to $1.6 billion, and free cash flow in the latest quarter was about $40 million. The full-year guidance depends on a large second-half ramp. China is 16% of sales by country of manufacture.

The strike question. A put about 18% below the late-September price sits near 21 times the midpoint of adjusted earnings guidance for the year to March 2027. Assignment held into 2027 would leave you with two stocks: a slower contract manufacturer and a concentrated AI power business. An investor day on 10 November is due to set longer-term targets.

Intel: profitable products, a loss-making foundry

Intel had $52.9 billion of revenue in 2025. Revenue in Q2 2026 was $16.1 billion, up 25%.

Operating income ($bn)	2024	2025	Q2 2026
PC chips	+11.6	+9.3	+2.3
Data centre chips	+1.4	+3.4	+2.5
Foundry	−13.3	−10.3	−2.1
GAAP gross margin	32.7%	34.8%	40.4%
Operating margin, GAAP / adjusted	−22.0% / −0.5%	−4.2% / 5.5%	11.1% / 17.2%

The product businesses are solidly profitable. Data centre chips ran at a 40% operating margin in Q2. The foundry lost $2.1 billion in the quarter. The investment case turns on whether the foundry reaches breakeven.

The bull case. Intel's 18A process is in high-volume production and its next process, 14A, is committed for volume in 2028. Fuller fabs lower the cost per wafer. Data centre revenue grew 59% in Q2 on demand for CPUs that run AI agents. Intel has strategic investment from the US government, NVIDIA and SoftBank.

Where it inverts. External foundry revenue was $293 million in Q2. Most of the foundry's reported growth is Intel buying from itself. A rising share price also produces GAAP losses: the liability tied to the government's escrowed shares cost $12.5 billion in Q2 and turned GAAP earnings per share to −$2.16. The foundry breakeven target is the end of 2027, and the CFO said in August it could slip into 2028.

The bear case. Weighted shares outstanding rose 17% in a year, and Intel sold another $20 billion of stock at $95 in August. The government, NVIDIA and SoftBank bought at $20 to $23 a share. Capex is guided above $20 billion a year, and Intel expects PC demand to fall by low double digits in 2026.

The strike question. A put about 17% below the late-September price sits at about 50 times forward earnings. Assignment would mean owning a turnaround already priced as if it works, with a large block of cheaply acquired stock sitting below your cost.

Nebius: an AI cloud funded by contracts, debt and new shares

Nebius rents GPU capacity. The AI cloud is about 98% of revenue: $582 million in Q2 2026, up 454%, with annualised revenue of $3.0 billion.

	2024	2025	Q2 2026
Cost of revenue, excluding depreciation	48%	31%	23%
Depreciation and amortisation, % of revenue	84%	76%	45%
AI cloud segment adjusted EBITDA	−$129m	+$59m	+$286m (50%)
GAAP operating margin	−437%	−116%	−30%

The headline gross margin near 69% leaves out GPU depreciation, which is the largest cost. The cloud earns a 50% margin before depreciation and still loses money after it. From 2026, servers are depreciated over five years instead of four, which lowers that cost line.

The bull case. Contracted demand: Microsoft for $17.4 billion to $19.4 billion through 2031, and Meta for up to $27 billion, of which $12 billion is committed, starting in 2027. Nebius cites more than $40 billion of customer commitments and expects over $9 billion of customer prepayments in 2026, which help fund the build. Guidance for 2026 is $7 to $9 billion of annualised revenue by year-end and about a 40% adjusted EBITDA margin.

Where it inverts. The 2025 GAAP profit came from revaluing the company's stake in ClickHouse. Operations were still loss-making. Operating cash flow is largely customer prepayments. NVIDIA bought about 21 million shares at roughly $95 each, and the restriction on selling them ended on 11 September.

The bear case. Capex is guided at $20 to $25 billion against $3.0 to $3.4 billion of 2026 revenue. Dilution comes from several directions: an at-the-market share programme, $5.75 billion of convertible notes that convert at $313 and $325, and shareholder approval in August to issue up to 20% more shares without pre-emptive rights. One customer accounted for 83% of receivables at the end of 2025.

The strike question. A put about 22% below the late-September price sits below the average price of recent share sales. Assignment would mean owning a company whose share count is set up to grow.

What the four have in common

All four passed the price screen. All four had a story that changed once the segment note was read, and the margin walk is where it showed. TTM's margin gains are real and come from the cyclical segment. Flex's growth sits in a business with two dominant customers. Intel's foundry numbers include its own purchases. Nebius's gross margin leaves out its main cost.

None of this makes any of them a good or bad company. It changes what a put seller is agreeing to. A put is an offer to buy the shares at the strike. So the research has to answer the question an owner would ask, and it has to be done before the put is sold. It is the same reasoning as buying with a margin of safety, applied at the strike. For the wider AI picture these four sit in, see the AI buildout's second phase.

Where the screener fits

The first layer runs in the levelbox screener, across the whole universe: a quality score from financial statements, a risk read from price action, and a delta cap per name that drops when earnings land before expiry. How that cap is set is its own post. The research in this article is the layer on top, done name by name, for the few you are seriously considering.

Educational and analytical only, not investment advice. These four names are worked examples of our own research method at one point in time, not recommendations. Do your own research before acting on any of it. Figures come from company filings and releases as of late September 2026 and will change. This article was drafted with AI assistance; if you spot a mistake, tell us.

Common questions

What should I research before selling cash-secured puts on a stock?: Research the business you might be assigned as well as the option. Check how the company earns its operating income segment by segment, how its margins build from gross to operating, whether the popular bull story survives the latest filings, what the bear case is (customer concentration, capex against free cash flow, debt, dilution), and what valuation the put strike implies. A put is a standing offer to buy the shares at the strike, so the question is whether you would be glad to own the company at that price.
Why isn't a wheel screener enough on its own?: A screener works from prices, volatility, open interest and financial ratios. From those you can check that a put pays at least what the stock actually moves, that the chain is liquid, and that no earnings land before expiry. Segment notes and earnings calls are outside its inputs. In the four examples here, every popular narrative had at least one part that inverted once the filings were read. None of those inversions would show up in a screen.
What is a margin walk and why does it matter to a put seller?: A margin walk traces revenue down to operating income: gross margin, then operating expenses, then operating margin, on both a GAAP and an adjusted basis. It shows which segment actually makes the money and what the adjusted figures leave out. For a put seller it answers whether earnings are growing because the business is getting better or because a cost has been moved out of the headline number, which matters if you end up owning the shares.
How do I judge a put strike on fundamentals rather than delta?: Convert the strike into a valuation. Divide it by guided or forward earnings to get the multiple you would pay on assignment, note how far it sits below the 52-week high, and ask what you would actually own afterwards, for example two companies after a spin-off, or a share count that is set to grow. Delta tells you the chance of assignment. The multiple tells you whether assignment is a price you would choose.