Options research
XLP for straddle and strangle sellers: five years of big moves that came back
The consumer staples ETF is the calm corner of the market. Yet over the last five years, XLP kept running 7–15% in one direction and then turning around. A short straddle or strangle wants exactly that: moves that come back. I measured how unusual it was, against random versions of XLP's own returns, and what followed a move from the day a position was opened.
Key findings
- Since November 2021, XLP made 26 big turns, where chance gives 16 to 18. Only 0.5–2.5% of random versions of its own returns produced as many. Utilities (XLU) showed the same, the S&P 500 (SPY) did not.
- After a 10% move from the entry day, XLP gave half of it back in 84% of cases (random: 50%), and ran another 5% first in only 17% (random: 53%). Forty days after the move, half of it was gone on average.
- Big moves were also rarer: only 57% of entries ever saw a ±10% move within a year (random: 75%), and moves took longer to form.
- It is a recent pattern. Over 27 years, XLP's swings looked like chance, even though it was the most mean-reverting of 14 ETFs over two weeks to three months.
The pattern
The chart shows what XLP traders know from the screen: a run from 71 to 81 in early 2022, back to 69 by June; 76 to 90 between January and late February 2026, back to 81 within three weeks; 87 to 80 this autumn, then straight up to 90 in October. Measured with a fixed rule, XLP turned around 26 times in five years. The rule scales with XLP's own volatility, so in a calm phase a smaller move counts: the two smallest legs, 4.3% and 4.6% in 2024, came when XLP's usual two-week move was only about 1.9%, and were more than twice that. The median leg was 9%, the largest 19%. The latest run, from 80 on 1 October to almost 90, is still open: it becomes a turn only once XLP falls back by the threshold, currently about 7.5% from the high.
More turns than chance
Every price chart has turns, so the number alone means little. The question is how many a random path with the same moves would have. To answer it, I shuffled XLP's own daily returns into random order 200 times. Each version keeps XLP's fat tails and its calm and turbulent phases, but loses any order in the moves. Then I counted the turns in each version with the same rule.
XLP's 26 turns lie outside the range of the random versions; only 2% of them got as many. A market is not a pure coin flip, though: short runs of rising or falling days, and reversals the next day, are part of how prices move. So I repeated the test with blocks of 5, 10 and 20 consecutive trading days kept together, shuffling only the blocks. The result got stronger: the random versions averaged 16 turns, and only 0.5–1% reached 26. Utilities, the other defensive sector, came close with 24 (3.5%). Health care and the Nasdaq 100 were within chance, and the S&P 500 had slightly fewer turns than its random versions, also within chance. Small turns of about one monthly move were normal for XLP (33 against 36): what came too often were the big swings.
What followed a move from the entry day
For a straddle or strangle seller, the turns matter from the day the position is opened. So I opened a position on every trading day since November 2021 and followed XLP's close until it was 5%, 7.5%, 10% or 12.5% away from the entry, up or down. From that day on, the question is: does the move come back, or does it keep going?
From 7.5% on, the difference is large. After a 10% move, half of it came back in 84% of cases, where a random path gives 50%. The move ran another 5% before coming back in only 17%, where a random path gives 53%. Forty days after a 10% move, XLP had given back 51% of it on average; a random path would have moved on by 3%.
| Move | Reached | Days | Half back | Days | Fully back | Days | After 40d |
|---|
Random versions in brackets. Reached: share of entries whose close moved this far within a year (250 trading days). Days: median trading days from entry to the move. Half back / fully back: share of moves that gave back half / all of the distance within 120 trading days, and the median trading days it took. After 40 days: average change 40 trading days after the move, in % of the move (negative = given back).
Two more things stand out. Big moves were rarer than chance: only 57% of entries ever saw a 10% move within a year, against 75% for a random path, and the moves that did happen took longer to form. And the counter-moves were not faster than random: half the move came back in a median of about four to six weeks, the full move in about two to three months. The difference was how often they came back, not how quickly.
Only the last five years
XLP does not always behave like this. Over the 27 years since 1999, it was the most mean-reverting of 14 sector, index and metal ETFs: its moves over two weeks to three months were smaller than its daily moves added up to (variance ratio 0.83 over two weeks and 0.68 over three months, both statistically significant; SPY 0.86 and 0.77, silver 1.11 and 1.19). But its big swings over those 27 years looked like chance: 3.5 turns a year against 3.7 for random versions, and the share of moves fully reversed by the next one was the same (51%).
Big falls were bought back over the whole period, as in most stock ETFs. What made the last five years different was that the rallies came back as well, and that the swings kept coming. From October 2023 to September 2024, XLP rose from 66 to 84 in a mostly clean trend, the exception within the five years. A pattern like this can end at any time.
What it means for straddle and strangle sellers
For a short straddle or strangle on XLP, the last five years were a friendly environment on the price side: big moves were rarer than chance, and when one came, the odds favored getting a large part of it back within one to two months, while running much further first was the exception. A seller who held through a 7.5–10% move against the position was more often relieved than hurt.
Prices alone do not decide a straddle; the premium does too. Priced at XLP's own daily volatility, a two-month straddle charged for moves about 14% larger than XLP actually made over the last five years. Over one month there was no such gap. A different gap, between option prices and the moves that followed, showed on single SPY trading days: SPY closed beyond one VIX-implied daily move on 18% of days, where the option price assumed about 32% (0DTE odds on SPY). Whether a given position paid still depends on the implied volatility at entry, which this study does not have for the past.
Two more limits belong next to that. First, the five-year window was chosen after seeing the chart, which makes the 2% look somewhat better than it is; utilities showing the same pattern makes pure chance less likely, but it is no proof. Second, five years hold only about 8 to 12 independent big moves per size, and daily entries overlap heavily. The numbers describe what happened; they do not promise that it continues.
Data and notes
- Daily closes of XLP, XLU, XLV, QQQ and SPY from Yahoo Finance, since 1998–1999, data through 30 October 2026. The five-year window starts on 15 November 2021.
- Turning points: a move ends when the price reverses by at least 1.5 times its usual monthly move (the volatility of daily returns with a 42-day half-life, times the square root of 21), counted on dividend-adjusted closes.
- Random versions: each ETF's daily returns, divided by its volatility on that day, shuffled into random order and rescaled with the same volatility path. 200 shuffles for the turn counts, 20 for the entry statistics. Block version: runs of 5, 10 or 20 consecutive days kept together and drawn at random, 200 versions each (XLP 0.5–1.0%, XLU 1.5–2.5%, XLV, QQQ and SPY within chance).
- Moves from the entry day: closing prices without dividend adjustment, as options settle on the price; a move counts on the first close beyond the distance, up or down.
- Straddle pricing check: average absolute close-to-close move over 21 and 42 trading days, divided by what the daily volatility at entry implies for that horizon (volatility × √days × √(2/π)). XLP since November 2021: 0.99 over one month, 0.86 over two months; since 1999: 0.86 and 0.83.
- Variance ratios and the 27-year comparison: 14 ETFs (9 sector ETFs, SPY, QQQ, IWM, GLD, SLV), daily returns divided by their volatility.
- All numbers count what happened in the past. They are not probabilities for the future, not forecasts and not financial advice.