One NVTS 16P. Seventeen dollars of premium. Assigned two days later, cut fifteen days after that at $10.94 — a $491.76 loss, or 29× the premium.

That is not bad luck. That is the default payout of the structure: you win a few dollars, and you lose dozens of times that.

1. What I Used to Believe

When I started selling puts, my logic sounded like this:

The stock won’t fall to the strike, so I collect the premium for free. And if it does fall, I wanted to own the stock anyway.

Both sentences are true. Together they miss one thing: on assignment day you pay the strike, not the market price.

On June 30 I sold one NVTS 16P with two days to expiry for $17. On July 2 I was assigned at $16 — the stock closed at $14.46 that day, so the position was underwater the moment I owned it. On July 17 I sold at $10.94. The ledger says −$491.76. Over those two months the option leg had earned +$195.

Options pay in pocket change; stock loses in whole notes. The structure was set on day one. You just can’t see it while the rent is coming in.

2. A Limit Order With a Waiting Fee

Strip the jargon. Selling a cash-secured put does exactly one thing:

You promise the market that if the price is below the strike K at expiry, you will buy 100 shares at K. In exchange, the market pays you P today.

So the only real parameter is the effective purchase price: K − P.

The QUBT trade: strike $8, premium $0.49, effective price $7.51. The stock was around $9.50, so I had placed a 25-day limit order at $7.51 that paid me to wait.

Compared with a normal limit order, it costs you three things:

• Cancellation: a limit order can be pulled for free; a short put must be bought back at whatever the market asks.
• Cash: a limit order only costs money when it fills; a short put locks up K × 100 immediately.
• Obligation: a limit order is a right; a short put is a duty. If it expires in the money, you must buy.

The third one is where all the risk lives. A limit order is a right; a short put is an obligation. The premium is the price of that obligation — not a rent cheque.

3. The Win Rate Is Designed; the Payout Is Not

This is the gap between buyers and sellers:

• Win rate can be tuned. The further out you sell the strike, the more often it expires worthless. Delta is a rough proxy for the probability of expiring in the money (a rule of thumb, not an identity): a 0.2-delta put expires worthless roughly 80% of the time.
• Payout cannot be tuned. The maximum gain on a CSP is the premium. The maximum loss is K × 100 minus the premium — all of your cash if the stock goes to zero.

The QUBT trade’s payout was $98 : $1,502, about 1 : 15.3. One zero equals 15.3 full rent cheques. It doesn’t take a zero: at $5 the loss is $502, about 5× the premium.

Stack position size on top: $1,502 of worst-case loss against my October 27 net worth of $6,357.77 is 23.6% — a single trade eating the entire per-name position limit (the 2% rule and sizing).

So evaluate a CSP by worst-case loss as a share of net worth first, and by annualized yield second. Reverse the order and you are using the win rate to advertise the payout.

4. Four Ledgers: the Denominator Is a Choice

Same trade, different denominator, and the annualized number moves by 5×. The QUBT trade: 25 days, $98 gross premium ($96.50 net of commissions).

Denominator	Value	Annualized
Cash secured (K × 100)	$1,600	89.4%
Margin requirement (20%)	$320	447.1%
Effective cost (K − P) × 100	$1,502	95.3%

(20% is an illustration; brokers and tickers differ. At 30%, that row becomes 298%.)

The fourth ledger is realized P&L, which only exists on the day you close. That explains a common self-deception: unrealized gains are not income, and premium received is not profit — account equity is a valuation until you close.

All three annualized figures are arithmetically “right,” but only one is usable: the cash-secured one. The 447% margin number is not a return; it is leverage dressed as a return — and it is the number most broker apps show you.

The deeper problem: 89.4% is not an expectation either. It is 6.1% over 25 days multiplied by 14.6, which assumes you can repeat it every month. In reality, one assignment takes that cash out of the game for months.

The full definitions of all four ledgers (cash, account, realized, coverage) live in Premium Is Not Income (in Chinese). One rule here: any annualized figure must state its denominator, its day count, and that it is annualized. Miss one and the number doesn’t exist.

5. The Real Edge Is IV Minus RV

At this point the question is sharp: if the premium is fairly priced, a CSP is a zero-expectancy game with a fat left tail — a higher win rate doesn’t change the expectation.

Sellers have exactly one systematic edge: implied volatility is, on average, higher than the volatility the underlying actually realizes. That spread is the variance risk premium, and the rent you collect is its monetization.

So “is this worth selling?” is not about how fat the premium looks. It is:

1. Is IV expensive relative to this name’s normal realized move?
2. Was the premium inflated by an event? Earnings, M&A, index changes push IV up and then collapse it — often on the same day it gaps through your strike.
3. Can this name’s historical range be covered by the premium? Fat premiums on high-vol small caps are the price of risk, not a discount.

Conversely, selling puts when IV is below RV is paying someone to take insurance from you. Not selling is better than selling wrong.

6. Three Parameters, Decided Together

Strike, expiry, and size are not three independent choices. They share one constraint: how much money you are willing to turn into how many shares.

① Strike: not “0.3 delta,” but “I would own it here.” Willing to hold long term → you can sell closer to spot and collect more. Only renting, never owning → sell further out and accept a thinner premium. Never willing to own → don’t sell puts. That is an identity question, not a strategy question.

② Expiry: 25–45 days, and away from earnings. Time is the seller’s friend, but only in the last month does it run fastest. Too short and gamma dominates; too long and your cash is locked while theta crawls. No new positions inside two weeks of earnings; if you already have one, shorten or buy it back — that is discipline, not prediction (earnings and event risk).

③ Size: real cash, not buying power. One line: contracts = available cash ÷ (K × 100), rounded down. Margin is the broker’s risk budget, not your money; positions built on it get margin-called in a drawdown, and the one thing a CSP must never do is force you to close at the low (margin is not your money).

Pass all three, then go back to section 3 and recompute the worst case. Above 20% of net worth, cut a contract.

7. Four Decision Points Before Expiry

You don’t need to watch the screen daily. You need to decide at four moments:

① 50–70% of max profit: close or hold? Don’t memorize a number. Use one rule: if the remaining premium is less than the expected value of a fresh trade, close and redeploy. The premium left is what you pay for “one more leg”; whether it’s worth it depends on the next opportunity (closing early buys freedom).

② Below the strike: take assignment, close, or roll? Price out all three worst cases. A roll is not free: it converts this loss into a longer lock-up plus more cash committed. The risk isn’t gone; it’s deferred and upsized (rolling trades a loss for time).

③ Earnings or events: shorten or leave? IV rises into the event and collapses after. Selling before it collects the most expensive rent and carries the largest gap risk. A small account cannot absorb a −30% gap.

④ Assignment day: execute the script. Assignment is not an accident; it is the other half of the contract. Write three lines before you open: will I take it, where is my stop, how long do I wait. I wrote none of them on NVTS, so “let’s wait and see” decided for me — fifteen days and $352 worse.

8. Three Ways This Actually Loses Money

Teaching pieces love the upside. Here is the other side. There are only three paths to a loss, and all three have happened to me:

One: a gap through the strike. A CSP has no stop, only assignment. Friday close $8.20, Monday open $6.50 — 200 shares down $300 while the $98 premium covers a third of it. This is a tail, not a probability question. It’s a sizing question.

Two: assignment, then more downside, with the cash already spent. The moment you’re assigned you go from “collecting rent” to “waiting for a bounce.” If you have a second and third put on the same name, you average down all the way to the bottom. “Keep buying as it falls” sounds like discipline; in practice it is doubling down.

Three: no liquidity, no exit. Wide bid-ask on small caps can eat the entire premium: you collect $0.49 and pay $0.35 to close — $0.14 is the market’s toll. In stress there may be no bid at all. Being able to close matters more than how much you collected (where to hang your order).

One more that gets ignored: early assignment. American options can be exercised any day, most often when deep in the money or just before a dividend. You think you have 20 days; you may own the stock tomorrow.

9. When This Framework Doesn’t Apply

1. The name is a core position you plan to keep adding to. Assignment is the plan, not an accident — you can relax the 20% line, but the cash must actually be there, not borrowed.
2. You run portfolio margin with several low-correlation names. Single-trade worst cases don’t simply add, so portfolio risk is lower — at the cost of larger positions and a much faster drawdown when correlations go to 1.
3. Very far out-of-the-money puts. Ugly payout, tiny probability, possibly positive expectancy — provided the position is small enough that zero doesn’t matter. That is not a recommendation; it is a reminder not to treat a high win rate as a moat.
4. You never look at IV. Mechanical monthly selling eventually gets hit in one earnings season. That isn’t the framework failing; it’s the process failing.

10. Seven Questions Before You Sell

1. If this put is assigned, do I want 100 shares at K? (No → stop.)
2. What is the worst-case loss in dollars, and as a share of net worth? (>20% → cut size.)
3. Is the cash real cash, or buying power?
4. Is there an earnings date or major event before expiry?
5. What is the annualized premium, and which denominator did I use?
6. Is IV expensive or cheap relative to this name’s realized moves?
7. If it drops 20% tomorrow, do I take it, close it, or roll it? Write it down now.

If you can’t answer all seven, you are not trading a strategy. You are hoping.

11. One Line

You can design the win rate on a short put. You cannot negotiate the payout.

So the order is always: worst-case loss first, premium second; IV first, fat premium second.

---

Related: Rolling is not a fix · Selling into earnings · Margin is not your money · Where to hang your order

Data & Sources

• Updated 2026-11-13 | First published 2026-11-13
• QUBT: 2× 18SEP26 $8 puts at $0.49 (filled 2026-08-24, ~$96.50 net, $1,600 cash secured, 25 days), from the trade review
• NVTS: 1× 16P at $0.17 (2026-06-30), assigned 7/2, sold 7/17 at $10.9353, realized −$491.76 per broker records, see the assignment review
• Account reference: TWS snapshot 2026-10-27 (net worth $6,357.77, cash $1,992, maintenance margin 49.4%)
• Annualized figures are arithmetic illustrations (gross premium ÷ denominator ÷ days × 365), not return expectations
• The delta-to-probability mapping is a rough approximation, not an identity

Risk Boundaries

This article covers structure and parameters. It is not investment advice and promises no return. A CSP can lose the entire cash secured: gaps, earnings, liquidity holes, and early assignment all defeat a plan. A structure can tell you your worst case; it cannot pay it for you.