Correlation risk: different tickers, shared exposure
Correlation changes joint outcomes even when individual probabilities stay fixed. Stress the whole group rather than multiplying marginal odds blindly.
By OptionsKing. Publisher and editorial methods.
What you will learn
By the end of this lesson, you should be able to:
- Distinguish marginal and joint probability.
- Explain clustering under dependence.
- Stress common exposure and delivery together.
Hold the three puts' individual modeled probabilities constant and vary only their dependence. The table then separates the chance of any assignment from the chance of all three. Its Gaussian factor model is an illustration, not a measured forecast for semiconductor stocks.
The three positions
From cycle 1 of the running book, all three semiconductor puts, 46 days out.
| Ticker | Stock | Strike | Implied volatility | Odds of assignment |
|---|---|---|---|---|
| AMD | $118.00 | $106 | 45% | 26.6% |
| MU | $92.00 | $83 | 42% | 25.7% |
| INTC | $23.40 | $21.50 | 38% | 27.4% |
Multiply those three together and you get 1.88 percent. That is the number a spreadsheet gives you, and it is wrong, because it assumes the three stocks do not know about each other.
What correlation does to the joint odds
Model the three as one common factor plus three independent wobbles, which is the standard way and is honest about what it assumes. Turn the correlation dial and watch two things move in opposite directions.
| Pairwise correlation | All three assigned | At least one assigned |
|---|---|---|
| 0.00, independent | 1.88% | 60.5% |
| 0.50 | 8.17% | 48.5% |
| 0.70 | 12.09% | 43.0% |
| 0.80, roughly a sector basket | 14.64% | 39.8% |
| 0.90 | 18.04% | 35.7% |
Two readings, and the second one is the one nobody expects.
Disaster gets eight times more likely. All three assigning goes from a 1-in-53 event to a 1-in-7 event. Nothing about any individual position changed. The strikes are the same, the deltas are the same, the premiums are the same.
And quiet quarters get more likely too. At least one assignment falls from 60.5 percent to 39.8 percent. Correlation does not make bad things happen more often. It bunches them. Six calm cycles then one that hurts, instead of a steady drizzle.
That combination is what breaks people. The bunching teaches you that the strategy is safe, right up until the cycle where it is not.
Three tickers, 1.1 bets
The same effect, measured on the share exposure rather than the assignment odds.
| Correlation | Basket volatility | Effective independent bets |
|---|---|---|
| single name, weighted average | 43.1% | 1.00 |
| 0.00 | 28.3% | 2.31 |
| 0.50 | 36.5% | 1.40 |
| 0.80 | 40.6% | 1.13 |
| 1.00 | 43.1% | 1.00 |
At 0.80, spreading across three semiconductor names cuts the volatility of the basket by 5.8 percent against holding just one of them. Three tickers, three commission bills, three sets of earnings dates to check, and 1.1 independent bets.
You did not diversify. You tripled the paperwork.
What the book actually did
Cycle 3 landed on the 14.6 percent branch. All three semis assigned, and so did WBD and KMI, so five of six positions turned into stock inside the same 46 days.
The individual moves were not extreme. AMD fell 1.67 standard deviations, which happens. The improbable part was the simultaneity, and simultaneity is exactly what correlation prices.
Note also what the calm cycles looked like. Cycles 1, 2, 4, 5, 6, 7 and 8 produced zero assignments between them. Seven quiet cycles and one that took the account down $2,902. That is the bunching, on the page.
Three things to do about it
Correlation is not constant, and a return correlation does not completely describe extreme joint outcomes. Staggered dates change delivery timing but do not remove a shared shock. Check the dollar loss and funding needs when the entire group is tested together.
- Cap the sector, not just the ticker. Any correlation above about 0.7 means you should treat the group as one position for sizing. The weights are the first place to look.
- Stagger the expiries. Every position in this book shared one 46-day cycle, so one bad six weeks caught all of them. Splitting across two or three expiry dates does not reduce correlation between the stocks, but it does stop a single window deciding the whole book.
- Size for the bunch, not the average. Plan the account around the 14.6 percent cycle rather than the 1.9 percent one. If the all-at-once outcome is survivable, the rest takes care of itself.
One thing not to do: assume ETFs solve it. An index put is a single position whose correlation with itself is 1, and cluster F works through what that trade is and is not. Diversification inside the underlying is real. It is not the same as diversification across your positions, and it does not stop everything falling at once.
Check your understanding
Why can higher correlation lower the chance of any assignment while raising the chance of all three?
Show the explanation
The outcomes become more clustered: more periods when all avoid assignment and more when all are assigned. That is possible while each individual assignment probability remains unchanged.
Common questions
How does correlation affect a portfolio of short puts?
In the specified simulation, higher correlation increases the chance of joint losses. The 0.80 value and assignment frequencies are model assumptions/results, not observed sector facts, and correlation alone does not specify the full joint distribution.
Is selling puts on three stocks in the same sector diversified?
Several same-sector names can share substantial exposure. The example's 0.80 correlation produces little diversification under its model, but is not a measured correlation for every sector or period.
Why do several of my positions always go wrong at the same time?
Because correlation makes clustered failure much more likely than the individual probabilities suggest. In the worked year, seven of eight cycles produced no assignments at all and one produced five.
Does selling index options fix correlation risk?
It diversifies inside the underlying, not across your book. One index put is a single position, and if it is the whole account then the correlation of your book with itself is 1.
Sources
Rules and thresholds above were checked against these documents on August 4, 2026. Exchange and broker rules change. Confirm anything you are about to act on with your own broker.
Keep reading
Do the math on your own trade
Every price in this article is an illustrative worked example, not a quote. Read Running the book for the rest of the series, and the disclaimer before you act on any of it. Selling options carries real risk of loss, and the loss can be far larger than the premium you collected.