Does selling options only when IV is high actually work?

"Only sell premium when implied volatility is high" is the most-repeated rule in options income - it's baked into every IV-rank dashboard and half the trading courses. So we tested it the only way that settles an argument: a 20-year backtest. We sold 30-delta cash-secured puts on the S&P 500 four ways - every month, only when IV rank was high, only when IV percentile was high, and only when IV rank was low - and gave every version the same risk-free interest on its idle cash so the comparison is fair.

The verdict is a split decision: timing by IV rank made each trade richer and the ride much smoother - but it traded so rarely that total return fell below simply selling every month. "Sell when IV is high" is right about trade quality and wrong about total return. IV rank turns out to be a risk dial, not a return booster.

The scorecard (2006–2026)

Four ways to sell the same 30-delta cash-secured put on SPY. Idle cash earns a 2% risk-free rate in every version, so strategies that sit out are not unfairly penalized:

Read the top two rows together. Selling every month earned 5.89% a year. Selling only when IV rank was high earned 3.30% - less - despite collecting a fatter premium each time it traded. The high-IV filter won on one axis only: a -20.6% drawdown versus -28.7%, at roughly half the volatility.

Real fills confirm it - and the high-IV arm never had a down month

The scorecard above prices each put off VIX. We re-ran all four arms on real OptionsDX end-of-day SPY premiums, 2010–2023, selling at the bid/ask mid with strikes chosen by real delta - same flat 2% on idle cash. The verdict survives intact, and the high-IV arm comes out even cleaner:

Selling every month still wins on return (6.04% versus the best timing arm's 4.36%), exactly as the model showed - the cash drag of sitting out is too costly. But the real data sharpens the risk story: in this window the high-IV-rank arm never had a single losing month - a 0.0% drawdown at just 2.1% volatility - because it traded only 11% of the time (18 times across the 2010–2023 window) and always into a volatility spike, when the real premium was richest: 1.87% of collateral per trade, nearly double the 1.06% of an average month.

Two honest caveats on that zero drawdown. It's partly the window - the modeled run that includes 2008 shows the high-IV arm taking a -20.6% hit, so treat 0% as "this particular stretch," not a guarantee. And real skew is what lifted the high-IV arm above the other two filters here (it earned 4.36% versus 3.65% for low-IV and 3.61% for high percentile) - rich out-of-the-money premium in the spikes is real, and the model understated it. Either way the conclusion is unchanged: high IV rank is the lowest-risk arm, not the highest-return one.

Growth of $100,000, 2006–2026

The chart makes the cash drag visible: selling every month (gold) compounds steadily, while the high-IV-rank version (blue) flatlines for years at a time - those are the long stretches it sat in cash, waiting for a volatility spike that mostly didn't come.

Why "IV rank above 50" fires only 14% of the time

This is the crux, and it's about the metric, not the market. IV rank measures where VIX sits inside its 52-week low-to-high range. A single spike - one COVID, one 2022 - sets the "high" for a full year and drags the midpoint up with it. So on the great majority of ordinary days, VIX sits well below that midpoint and IV rank reads under 50. In our 20 years, IV rank was 50-or-higher only 13.9% of months.

Swap the metric and the behavior flips. IV percentile - the share of the last year's days that were calmer than today - crossed 50 about 45.9% of the time, more than three times as often. And here's the kicker: selling on high IV percentile (3.01% CAGR, -29.1% drawdown) was strictly worse than selling every month - it traded less, earned less, and got no drawdown benefit. If you're going to use a high-IV filter at all, the metric you pick decides everything (IV rank vs IV percentile explains the difference in full).

Each high-IV trade really is better - the catch is how few there are

The maxim isn't wrong about trade quality. When IV rank was high, each cash-secured put paid 1.83% of the collateral in premium, versus 1.13% on an average month and just 1.02% in the calmest (low-IV-rank) months. Richer premium also means a fatter cushion before assignment, which is most of why the high-IV-rank drawdown was so much shallower.

But a great premium on 34 trades couldn't out-earn a decent premium on 244. With idle cash earning only the risk-free rate, every month you sit out is a month your capital barely works. That's the whole story of the 3.30%-vs-5.89% gap: not worse trades, just far fewer of them.

What it means for how you sell

• If your goal is total return, selling consistently beat waiting for high IV. The premium you collect in ordinary months adds up to more than the richer premium you'd collect in the rare rich ones.
• If your goal is a smoother ride, the high-IV-rank filter delivered - a -20.6% drawdown at half the volatility. Use IV rank to size up when premium is genuinely rich and step back when it isn't.
• Mind your metric. "IV rank > 50" and "IV percentile > 50" are not interchangeable - one fired 14% of the time, the other 46%, with different results. Know which one your platform shows you. Check it on the IV rank calculator before you trade.

Caveats - read these

• Cash earns 2% here, on purpose. Unlike our other backtests (which use a conservative 0% on idle cash), this study credits idle and collateral cash a 2% risk-free rate in every strategy - because the entire point is comparing approaches that sit in cash for very different fractions of the time, so they must be treated identically. That lifts the absolute CAGRs a little above our other put-selling figures; what matters here is the gap between the rows, not the level.
• The scorecard prices premiums from VIX; the real-fill section uses traded prices. The headline 2006–2026 table is modeled from VIX (a conservative floor, since it understates the skew of real out-of-the-money puts). The "real fills" section above re-runs all four arms on actual OptionsDX premiums (2010–2023) and lands in the same place - always wins on return, high IV rank is the low-risk arm. The ranking is robust either way.
• One window, one underlying, one threshold. SPY, 2006-2026, a 50 cutoff, held to expiration with no management. A different threshold, a single stock, or a sideways decade could shift the numbers - though the structural point (high IV rank is rare, so timing means sitting in cash) holds.
• Past performance is not predictive. Educational, not advice.

Method: real daily SPY and VIX, 2006-02-17 to 2026-05-15 (244 monthly third-Friday cycles after a 252-day warmup). 30-delta cash-secured puts priced with the same Black-Scholes engine as this site's calculators; idle/collateral cash earns a flat 2% risk-free rate in all strategies; held shares on assignment marked to the expiration close. IV rank = (VIX - 52-week low)/(52-week high - 52-week low); IV percentile = share of the trailing 252 days below the current VIX. The "real fills" section re-runs the identical four arms on real OptionsDX end-of-day SPY chains (2010–2023), every put at the bid/ask mid with strikes by actual delta (IV rank still from VIX) - independently re-derived to the dollar. Every figure regenerates from the underlying data; none are hand-entered.