Why we asked
In this week's Deflow research note, the strategy's biggest losses came from bullish put spreads on both SPY and QQQ. Its risk rules limit exposure per symbol, so the two positions were sized as if they were separate risks. We wanted to know how separate they really are.
What the data shows
| Pair | Correlation | Lowest 20-day | Highest 20-day |
|---|---|---|---|
| SPY – QQQ | 0.93 | 0.82 | 0.99 |
| SPY – IWM | 0.82 | 0.49 | 0.95 |
| QQQ – IWM | 0.74 | — | — |
Correlation of daily close-to-close returns, adjusted for dividends. 20-day figures are rolling windows over the same year.
They fall together without exception. SPY fell more than 1% on 26 days in the past year. QQQ fell more than 1% on every one of them, and IWM fell too. On those days SPY averaged −1.41%, IWM −1.71% and QQQ −2.01%.
The relationship does not drift much. Measured over any 20-day window, the SPY–QQQ correlation never dropped below 0.82. SPY and IWM are more loosely linked: their 20-day correlation ranged from 0.49 to 0.95.
QQQ moves further. QQQ's annualized volatility over the year was 19.8%, against 12.7% for SPY. On SPY's down days QQQ fell about 1.4 times as far. A QQQ spread with the same strike distance as a SPY spread is therefore the riskier of the two.
Two positions, about one bet. A common rule of thumb for equal-sized positions is N ÷ (1 + (N − 1) × ρ), where N is the number of positions and ρ their correlation. For SPY and QQQ at 0.93, that gives about 1.04 independent bets. Adding IWM to the pair only raises it to about 1.13.
Last week as an example
From the October 9 close to the October 16 close, QQQ fell 5.5%, IWM 3.1% and SPY 2.0%. All three fell, and QQQ fell more than twice as far as SPY. QQQ finished at $709.80, below both strikes of the 725/710 put spread Deflow had sold. SPY finished at $763.25, still above its short strikes. The SPY and QQQ positions were the same bet, and the QQQ one was the larger version of it.
What it means for risk limits
- Per-symbol caps can understate risk. A rule like “3% of equity per symbol” allows 6% on SPY and QQQ combined, even though they rarely diverge on bad days.
- Grouping correlated symbols is one fix. SPY, QQQ, XSP, SPX and DIA all track large-cap US stocks. A limit applied to the group, not each ticker, matches how they behave in a sell-off.
- Beta-weighting is another. Expressing every position in SPY-equivalent terms, scaled up for QQQ's larger moves, gives one number for how much the book loses if the market falls.
Limitations
- One year of data. Correlations change across regimes. The past year had only two days with SPY down more than 2%.
- Returns, not option P&L. Credit spreads lose money non-linearly as price approaches and passes the short strike. Two spreads on correlated underlyings can still behave differently if their strikes sit at different distances from the price.
- Close-to-close only. Intraday moves, which can trigger stops, are not captured.
Open question
What is the right way to limit correlated short-premium exposure in an automated strategy: correlation groups, beta-weighted delta, a stress test of the whole book, or something else? We would like to hear how others handle it in the Community Discussion.
Price data from Yahoo Finance. This note is research and education, not investment advice or a recommendation to trade any strategy or security. Past performance does not guarantee future results. Options involve substantial risk and are not suitable for all investors.