Selling far-OTM puts for safe income is a myth

The pitch is seductive: sell a put so far out of the money it almost never gets hit, pocket the premium, repeat - "safe" income with a 90-95% win rate. It is the most popular cash-secured put idea on retail forums, and on real data it is a trap. Over 14 years of actual SPY option fills, the far-out-of-the-money 10-delta put made just 1.31% a year - barely above cash - and in the one crash that mattered it lost more than selling at the money. Its 95% win rate hides a near-zero return.

This is not a model. We ran the same monthly cash-secured put on real OptionsDX end-of-day SPY option chains, 2010–2023 (163 cycles): at each roll we pick the put whose actual delta is closest to the target and sell it at the real bid/ask mid, fully cash-secured, held to expiration. Here is every strike, same window, same rules.

Real OptionsDX SPY fills, 2010-01-15 to 2023-12-15, sold at the bid/ask mid with strikes chosen by actual delta; collateral held fully in cash, idle. "Win rate" = the put expired worthless and you kept the premium - not that the year was profitable.

The win rate is a magic trick

Look at the 10-delta row again: it won 95% of its months - it kept the premium almost every single time - and still turned $100k into only $119k over 14 years. How? The premium was so thin that the handful of months it did get assigned, deep below the strike, gave back most of what the winners ever collected. Many tiny wins, a few real losses, a near-zero net. You collect a fraction of a percent a month to stand in front of a 25% hit - win ninety-five of those and lose five, and the five still own you.

That is the whole illusion in one line: win rate measures how often you win, not how much. A 95% win rate on pennies loses to a 77% win rate on real premium - which is exactly what the at-the-money strike did, compounding 6.03% a year ($222k), about 4.6 times the far-OTM return, off nothing but a different strike.

The crash inverts the whole pitch

"Far-OTM is safer" should at least hold up when the market falls apart. It does the opposite. Here is 2020 - a sharp crash and a fast recovery, the worst case for any option seller - by strike, from the same real fills:

The "safe" strike lost the most. The mechanism is simple once you see it. The 10-delta seller collected a few cents, then watched SPY fall roughly a third straight through the strike - the premium cushioned almost nothing, so the loss was nearly the full move. Worse, there was no fat premium to rebuild with: after the crash, volatility stayed high for months, and the at-the-money seller banked rich premium every cycle that clawed most of the loss back (-6.1% on the year). The 10-delta, collecting pennies, could not - it finished -19.8%. The at-the-money put even took the deeper single-month hit in March (-29.6% versus -25.1% far OTM), yet still finished the year far ahead - the premium that cushions the blow is the same premium that funds the recovery.

So far-OTM is the worst of both worlds: the least income in the calm years, and - when it finally matters - a bigger hole than the strike everyone calls "risky," with no premium left to climb out of it. The premium you skip is the cushion you don't have.

It is not an SPY fluke

The same shape shows up on QQQ over 12 years (2012–2023): the far-OTM 10-delta made 2.42% a year to the at-the-money strike's 7.82%, and in 2020 the 10-delta lost 11.1% while at-the-money actually finished +4.9%. Two underlyings, two windows, one conclusion: the closer to the money you sell, the more you make, and far-OTM is not the safe harbor it looks like.

Why the myth is so sticky

It survives because it is built to feel right. A 95% win rate produces a long, smooth string of green months, so it feels safe and consistent - right up until the rare month that erases a year of them. The losses are infrequent and lumpy, so they never make it into the screenshots. And "10-delta = ~90% chance it expires worthless" gets mistaken for "~90% chance the trade is a good idea," which is a different claim entirely. The probability of keeping the premium is high; the amount is tiny and the tail is real.

What to do instead

Selling cash-secured puts is a perfectly good income trade - just not at the strike most people reach for. The real fills point one way:

• Sell the income strike, around 30-delta - meaningful premium without going all the way at-the-money. It carried its weight (3.51% a year) where far-OTM did not. The wrong reason to pick a strike is "it almost never gets hit"; the right one is "the premium actually pays me for the risk."
• Only on a stock or ETF you would genuinely be happy to own at the strike. The premium is a bonus for taking a position you wanted anyway - not a reason to sell on a name you would never hold.
• When the premium is genuinely rich, not just present. Check it with the IV rank calculator - high IV rank is when selling pays.
• Sized to a real dollar yield. Run the trade through the cash-secured put calculator: if the annualized return on capital is low single digits with a huge cushion, you are in the 10-delta trap.

The one rule that falls out of the data: never sell the far-OTM "safe" strike for income. It is the one that barely beat cash and got hit hardest in the crash. Safety in put-selling comes from the stock you choose and the size you take - not from a strike so far away it pays you nothing to stand in front of the same risk.

Caveats - read these

• Idle collateral. The backtest leaves the cash collateral earning nothing, so the 10-delta's 1.31% is mostly premium on top of idle cash. Parking that cash in T-bills would lift every strike - but equally, so the gap between far-OTM and the income strike, and the crash result, is unchanged.
• One window. 2010–2023 on real option data (it covers the 2020 crash but not 2008). The longer, modeled record in the full backtest tells the same story.
• Mechanical rules. Hold-to-expiration, no rolling, no profit-taking, no dodging earnings. Active management changes the numbers - but does not turn thin premium into a cushion.
• Educational, not advice. Past performance is not predictive.

Source: real OptionsDX end-of-day SPY and QQQ option chains (2010-01-15 to 2023-12-15), sold at the bid/ask mid with strikes chosen by actual delta, fully cash-secured and held to expiration. Every figure regenerates from the data; none are hand-entered. For the full strike ladder, equity curves and the 19-year picture, see cash-secured puts vs buy-and-hold.