Research
Delta is already the volatility adjustment
Selling a smaller delta on high-volatility names sounds prudent. Measured across volatility terciles, outcomes are flat — delta has already done that work.
Measured August 21, 2026
The idea that sounds right
Sell a smaller delta on high-volatility names. It feels obviously prudent: a name that swings 80% annualized is more dangerous than one that swings 15%, so give it more room.
We were about to build it. First we checked whether a constant delta was actually leaving high-volatility names less protected.
It is not
Delta is not a distance. It is a probability, and the option market computes it from the volatility of the underlying — so holding delta constant already widens the strike as volatility rises. It does so automatically, continuously, and using the market’s own estimate rather than ours.
At a constant ~0.22 delta, the cushion between spot and strike rises from 10.7% to 22.7% as implied volatility goes from 0.15 to 0.99. The adjustment we were about to add by hand was already happening.
And the outcomes are flat across the volatility spectrum. Splitting the history into three equal volatility buckets of roughly 2,400 windows each:
Breached at expiry: 18.5% / 18.6% / 19.4%. Traded through at any point: 39.0% / 40.6% / 39.5%. Low, middle and high volatility, and nothing to choose between them.
What building it would have cost
A smaller delta on high-volatility names would have double-counted volatility — once in the delta calculation, once in our own adjustment on top — and given up exactly the bucket where the high annualized returns live. It would have looked like prudence and behaved like a tax.
The companion idea, also shelved
The obvious next question is whether a particular ticker breaches more often than its delta implies. We cannot answer that with the history we have, and it is worth being precise about why rather than shipping a number.
Per-ticker residuals have a standard deviation of about 15 points across roughly 2.3 independent windows each. Resolving a 10-point residual would need something like 64 effective windows — four years of history per name at a 21-day managed hold. A validity check confirms the data is not there: the correlation between a name’s implied volatility and its empirical breach rate is 0.02, when higher-volatility names ought to breach a fixed cushion more often.
Pooled, the measurement is clean and flat: delta over-predicts breaches by 4–5 points at every volatility level. That is the variance risk premium. It is a fact about selling options, not a fact about any ticker.
What this does not show
This measures frequency, not severity. Delta equalizes how often a strike is reached across volatility levels; it does nothing about how bad the outcome is when it happens, and being assigned on a 3x leveraged fund is a far worse day than being assigned on a utility. The useful control there is position size or an outright exclusion, not the strike.
Every gate, factor, weight and curve behind this is on the methodology page, and every pick the model has published is on the track record.
Research and education, not investment advice. No result here is a forecast. See the disclaimer.