Strategy family 11 / 12
Options & volatility
Options strategies trade uncertainty itself: how far prices may move, how that expectation is shaped across strikes and maturities, and how heavy the tails might be. The family covers buying and selling volatility outright, relative value across the volatility surface, income overlays such as covered calls, positioning for convexity, and protection that ranges from tail hedges to option-based portfolio insurance. What unites them is a nonlinear payoff, so results depend on the size and timing of price moves, not only on their direction.
Covers
- Volatility relative value
- Long & short volatility
- Option income overlays
- Convexity
- Tail hedging
- Option-based portfolio insurance
Why it is a demanding test
Option payoffs are nonlinear and hedged positions are path-dependent, so a correct view on direction or volatility can still lose money through the wrong strike, tenor or hedging frequency. The family tests whether a system keeps its Greeks straight while the surface moves, and whether it resists positions that look steady until a single gap move. Option bid-ask spreads are often wide relative to premium, so cost awareness is part of the test from the start.
What we test
What a system needs to do well in this family. Tasks are built so that each ability can be observed and scored on its own.
Interpreting the volatility surface
Read implied volatility across strikes and expiries, including skew, smile, term structure and event premia. Explain what the shape says about demand for protection and expected risk, and how it tends to shift when spot moves.
Weighing realized against implied volatility
Estimate realized volatility with care over window length, sampling frequency, overnight gaps and jumps. Judge whether implied volatility is rich or cheap after allowing for scheduled events, such as earnings releases, and any risk premium built into option prices.
Keeping the Greeks straight
Track delta, gamma, vega and theta, along with second-order effects such as vanna and volga, across a whole book as spot and volatility move. Know when a book that is flat in total vega still carries skew or term-structure risk.
Designing delta-hedging rules
Decide when and how much to re-hedge, knowing that a hedged option's profit depends on the path of the underlying, not only on realized volatility. Each re-hedge pays a spread, so hedging frequency is a cost decision as well as a risk decision.
Structuring protection on a budget
Build tail hedges and portfolio insurance from puts, put spreads, collars and calendar structures under a fixed premium budget. Weigh the steady cost of protection against the drawdowns it addresses, and check the basis risk between hedge and portfolio.
Example task types
Illustrative designs that show the shape of the work. They are not items from the benchmark itself.
- Track: Analysis
Extract the implied move before an earnings release
The system receives a point-in-time option chain on a single stock ahead of its earnings date, along with the stock's past earnings-day moves. It must back out the implied event move from the term structure of implied volatility, compare it with the historical distribution, and describe defined-risk structures that could express a view. A good answer shows the variance arithmetic, notes the small sample and does not treat one comparison as an edge.
- Track: Simulation
Maintain a tail hedge on a fixed budget
In a multi-year historical replay, the system protects an equity portfolio under a fixed annual premium budget, choosing strikes, tenors and structures and deciding when to roll or monetize. Option prices come from the historical surface with bid-ask spreads, so every choice has a cost. Scoring weighs drawdown reduction per unit of premium across calm and stressed periods, and whether protection was in place when it was needed.
- Track: Live
Forecast realized volatility against the implied
Each week, on live data, the system forecasts next-month realized volatility for a fixed set of underlyings and states whether implied volatility looks rich or cheap. Because the outcome has not happened yet, the answer cannot be recalled from training data. Forecasts are scored over rolling windows, with implied volatility and simple historical estimates as baselines.
- Track: Client
Explain what a covered-call fund gives up
A retiree asks whether the high distribution yield of a covered-call fund is reliable income. The system must explain that the premium comes from selling upside while most of the downside is kept, distinguish distributions from total return, and relate both to the user's need for income and tolerance for losses. A good answer is plain and balanced, and neither recommends nor dismisses the product.
How it is scored
Each run is scored on its outcome and on the process behind it, so a lucky result is not mistaken for a sound one.
- 01
Greek-level P&L attribution
Results are decomposed into delta, gamma, vega, theta and an unexplained residual. This shows whether a position earned what it was designed to earn, and flags a volatility view that made its money from direction.
- 02
Volatility forecast loss
Forecasts are converted to variance and scored with a loss function such as QLIKE, which ranks forecasts consistently even when realized variance is measured with noise. Implied volatility and historical estimates serve as baselines, so the score rewards information beyond what option prices already contain.
- 03
Protection per unit of premium
For hedging strategies, we divide the reduction in stress-period drawdown by the net premium spent over the full period, calm years included. This rewards protection that is affordable to keep, not only protection that pays off once.
- 04
Tail-aware risk-adjusted return
Return is measured against expected shortfall and maximum drawdown, with skewness reported alongside, rather than against volatility alone. Short-option payoffs can look smooth for long stretches, and this measure does not take that smoothness at face value.
- 05
Staying within Greek and margin limits
A rule and rubric check covers whether the book stayed within vega, gamma and margin limits, and whether short-option exposure was sized to stress scenarios rather than recent calm. It also asks whether the system planned for margin to rise with volatility.
Diagnostics
Failure modes we diagnose
Recurring mistakes the tasks are built to surface. Each one is reported separately, so builders can see which to fix first.
- 01
Pennies in front of a steamroller
Selling options for steady premium and sizing on recent calm, so that a single gap move can erase years of collected income. We measure exposure against stress scenarios, not trailing volatility.
- 02
Mistaking direction for volatility
A long-volatility position that profited only because the market moved one way, or a hedged book that quietly drifted into a directional bet. Greek attribution separates the two.
- 03
Flattening the surface to one number
Treating implied volatility as a single figure, ignoring skew and term structure, or assuming the surface stays put when spot moves. Relative-value views depend on exactly how the surface reshapes.
- 04
Assuming fills at mid
Backtests that trade at mid on wide, thin option markets overstate results. We fill at quoted bids and offers with size limits, and model early exercise, assignment and pin risk around dividends and expiry.
- 05
Hedges that lapse too early
Protection that expires before the drawdown, references the wrong underlying, or is dropped after a costly calm stretch because its cost was never budgeted. We check whether the hedge existed when the risk did.
Environment and data
We use point-in-time option chains with bid and ask quotes, implied volatility surfaces, the dividends, rates and borrow costs known at each date, and the underlying prices used for hedging. Fills happen at quoted prices with spread and size limits, margin rises with volatility, and early exercise, assignment and settlement at expiry are simulated. Replays include volatility spikes, gap moves and long calm stretches, so long- and short-volatility approaches each meet the conditions that test them.