Options Calendar Spreads for Quants

A calendar spread sells an option at one expiry and buys an otherwise comparable option at a later expiry. It is commonly described as a time-decay trade, but the economic object is a view on forward implied variance, spot path, smile dynamics, and an event that may occur between the two maturities. The long back-month option normally has more vega; the short front-month option normally has more theta and gamma.

Forward variance is the starting point

For maturities T1 < T2, total implied variance is w(T)=sigma(T)^2 T. The variance implied for the interval is:

sigma_fwd^2(T1,T2) = [w(T2) - w(T1)] / (T2 - T1)

This is not simply sigma(T2) - sigma(T1). A downward-sloping IV curve can still imply a positive and expensive forward variance. Work in total variance, using matched delta points and consistent forwards.

Structure	Short leg	Long leg	Main thesis
Long calendar	near expiry	later expiry	front IV is rich or realised front move is contained
Short calendar	later expiry	near expiry	forward vol is cheap / event is underpriced
Diagonal	different expiry and strike	different expiry and strike	term structure plus directional or skew view

Strike selection is a risk choice

Same-strike calendars are not same-moneyness calendars after a spot move. A calendar struck near current spot is usually long vega and long back-month exposure, but it can lose sharply if spot moves through the short option before expiry: the near option develops high gamma and its realised variance overwhelms collected theta. A delta-based diagonal may maintain a desired moneyness view, but it adds a surface interpolation and rebalance rule.

def forward_variance(iv_front, t_front, iv_back, t_back):
    w_front = iv_front**2 * t_front
    w_back = iv_back**2 * t_back
    if w_back < w_front:
        raise ValueError("calendar-arbitrage or inconsistent inputs")
    return (w_back - w_front) / (t_back - t_front)

The validation is deliberately strict for a single matched point. Real surfaces can contain noisy quotes; smooth total variance with calendar-arbitrage constraints before producing a signal. See implied-volatility-surface for why interpolation itself is a trading risk.

P&L decomposition

For a delta-hedged calendar over a short interval:

dPi ≈ (Theta_back - Theta_front) dt
      + (Vega_back dIV_back - Vega_front dIV_front)
      + 0.5 (Gamma_back - Gamma_front) d<S>

The sign of net theta is often positive at inception for a long calendar, but it changes with spot and time. The position is not guaranteed to profit merely because time passes. A front-vol collapse can help the short leg, but a parallel vol collapse frequently hurts the larger-vega back leg. A front-specific crush is the desired outcome; a full curve repricing may not be.

Events and term-structure extraction

Earnings and scheduled macro events often create a front-expiry premium. Compare the option-implied move with a robust distribution of comparable historical moves, but separate ordinary days from event days. If the event lies before T1, a long calendar generally sells the event and keeps post-event exposure. If it lies between T1 and T2, it does the opposite.

Estimate event variance by subtracting non-event total variance:

w_event ≈ w_with_event - w_without_event

This decomposition requires similar maturities and a stable baseline. A change in skew, liquidity, or market regime can masquerade as event variance. Do not infer a precise “earnings move” from two stale mids.

Execution and risk controls

Trade calendars as multi-leg orders when possible. Legging can leave unhedged gamma during a rapid move. Mark with bid/ask-aware package prices, not independent mid prices, and include hedge costs. Predefine a rule for the short option near expiration: assignment, pin risk, dividend exercise, and after-hours moves are operational rather than theoretical concerns.

Stress a 1–2 standard-deviation spot move, a front-only IV crush, a parallel 5-vol-point move, a skew steepening, and a discontinuous post-event gap. Model all legs under the same Black-Scholes pricing convention, but do not mistake that model's smooth Greeks for actual gap protection.

Key takeaways

• Calendars are forward-variance and surface-dynamics trades, not simple theta trades.
• Infer forward volatility from total variance, not differences in quoted IV.
• Spot movement can make a nominally long calendar short gamma where it matters.
• Separate event variance from ordinary term structure with matched, clean quotes.
• Manage package execution, expiration operations, and term-structure shocks explicitly.