Strike selection: what delta and IV are actually telling you

“Sell the 30 delta” is the most-repeated rule in retail options, and it is not a rule. It is a coordinate — a way of saying “about this far out” that automatically adjusts for volatility and time. That is genuinely useful. It is also routinely mistaken for a probability, and on a high-volatility name that mistake is worth eight percentage points.

Four real chains, taken apart: what delta measures, where it stops matching the true probability of finishing in the money, why probability of profit is a different number again, and what implied volatility is actually charging you for.

Delta is not probability

Delta is the first derivative of the option’s price with respect to the underlying — how much the contract gains if the stock moves a dollar. In Black-Scholes, a call’s delta happens to equal N(d₁). The risk-neutral probability of finishing in the money is N(d₂). Those are different quantities, d₁ is always greater than d₂, and the distance between them is:

d₁ − d₂ = σ√T

So the delta-as-probability approximation is fine when volatility and time are small, and degrades exactly in the situations where you most want the number to be right. Here is the AAPL September chain, 48 days out, with the chain’s delta next to the engine’s true probability:

On AAPL the gap is one to two points. Delta as a rough probability is defensible; nobody is going bankrupt on that error. Now do it on names with different volatility:

Note the direction carefully. For calls, delta overstates P(ITM). For puts, |delta| understates it. If you sell puts and read delta as your assignment probability, you are systematically under-forecasting assignment — the exact error a put seller cannot afford.

In the money is not the same as losing

Third column, third concept. Go back to the AAPL table and take the $325 call:

• Delta 0.319 — the option gains $0.32 for every $1 AAPL rises.
• True P(ITM) 30.0% — three times in ten it finishes above $325 and gets exercised.
• P(profit) 76.1% if you sold it — because breakeven is $325 + $6.01 = $331.01, and AAPL has to clear that, not merely $325, before the seller is down a dollar.

The band between $325 and $331.01 is a six-dollar strip where the option is assigned and the seller still made money. Confusing those three numbers is how people end up believing they win 30% of the time on a trade that wins 76% of the time and loses badly the rest.

There is a fourth number nobody prints and everybody should: the probability of touching the strike at any point before expiry, which for a driftless process is roughly twice the probability of finishing beyond it. Around 60% for that $325 call. If you manage positions by closing when tested, that 60% is your real trade frequency — not 30%.

What implied volatility actually buys

Hold delta constant at 30, hold duration constant at 48 days, and vary only the underlying:

The last column barely moves. Across a 2.7× range of implied volatility and a 5× range of share price, the credit for a 30-delta put 48 days out is about 8.1% of implied volatility, times the share price, every time. PLTR pays three times what KO pays because PLTR’s implied volatility is three times KO’s. That is the entire explanation.

Which is why “sell high IV” is only half a sentence. High implied volatility means a bigger credit and a proportionally bigger distribution of outcomes. You are not being handed an edge; you are being offered a larger bet at roughly fair odds. The edge, if you have one, is in disagreeing with the market’s volatility estimate — and that is a much harder claim than “the premium looked good.”

The expected move, and where 30 delta sits

The cleanest translation from IV to something a human can picture:

The market’s central expectation is that AAPL spends roughly two-thirds of the probability mass inside a $63 band over the next seven weeks. The 30-delta $325 call sits 0.51σ above spot. That is the real content of “sell the 30 delta”: sell about half a standard deviation out.

Stated that way, the rule stops sounding scientific and starts sounding like what it is — a reasonable default with no particular claim to optimality. Half a sigma is close enough to collect a real credit and far enough that ordinary noise does not take you out. There is nothing sacred about it.

Time does not scale the way you think

Option value grows with the square root of time, not with time. Twice the days is about 1.41× the premium, not 2×. Here it is in live quotes, on the at-the-money strike where the relationship is cleanest:

AAPL lands within half a percent of the theoretical ratio. That is not a coincidence and it is not curve-fitting — it is the square-root-of-time law showing up unedited in a real chain, which is a decent sanity check that the pricing model is describing the market rather than the other way round.

KO comes in low, at 1.431×, and the reason is in the IVs: 22.3% for August against 21.2% for September. The term structure is downward sloping — the front month is pricing something the back month is not. When your observed ratio diverges from √T, you have found a term-structure signal, not a mispricing. Go and look at what is in the near expiry before you decide you are cleverer than the curve.

So how do you actually pick a strike?

Backwards from every article that starts with delta. The order that survives contact with a real account:

1. 1Pick the price you are willing to transact at. For a put: the price at which owning 100 shares is genuinely appealing. For a covered call: the price at which selling your shares would not annoy you. This is a judgement about the business, and it is the only step where you have any edge at all.
2. 2Look up what the chain pays for that strike. Not the other way round. If the credit is insulting, the answer is “no trade,” not “move the strike.”
3. 3Check the delta as a sanity gauge, not a probability. Under ~0.15 you are usually collecting too little for the tail you are carrying. Over ~0.45 you have taken a directional position and should admit it.
4. 4Check the true probability and the breakeven. The engine prints both; they are not the same number and the difference is your margin for being slightly wrong.
5. 5Check what is inside the expiry. Earnings, product events, index rebalances, ex-dividend dates. Unusually rich premium is almost always a calendar fact.
6. 6Size against the assignment, not the credit. Strike × 100 × contracts is the obligation. The credit is the small number.

Do it on the ladder, not in a spreadsheet

Every table here is a still frame of something the builder recomputes live. Drag a leg chip along the strike ladder and the premium re-quotes from the chain while the payoff curve, breakevens, probability of profit, and Greeks track the drag frame by frame. Ten seconds of scrubbing teaches the relationship faster than any static table can.

For the seller’s view of the same contract, open it as a covered call against 100 shares. To see the volatility argument rather than read it, put the PLTR 30-delta put in a second tab next to the KO one — same delta, same days, wildly different shapes. And if you would rather have candidates ranked than eyeball a ladder, the optimizer enumerates the real chain against a target price and date and sorts by return on risk.

More context per ticker on the AAPL covered-call page and the PLTR cash-secured-put page.

Caveats

Next: put this to work in the cash-secured put guide, or see what happens after assignment.