Volatility Calendar Spreads (Trading the Term Structure)

Overview

Options on the same underlying but with different expiry dates do not all trade at the same implied volatility. Plot implied vol against time to expiry and you get the term structure of volatility.

Most of the time it slopes upward: near-dated options are cheaper in implied vol terms than far-dated ones. That is the normal, calm state of the world, and it reflects the fact that uncertainty about the distant future is genuinely greater than uncertainty about next week.

But when the market gets frightened, this flips hard. Short-dated implied vol spikes far more than long-dated implied vol, because the fear is about right now, not about next year. The curve inverts. And historically, that inversion does not last. The market calms down, the front-month panic subsides, and the curve reverts to its normal upward slope.

The calendar spread is how you bet on that reversion. Sell the expensive front, buy the cheaper back, and wait for normality.

Strategy logic

• The signal: Measure the slope, front-month at-the-money implied vol minus back-month at-the-money implied vol. When that number is unusually high, meaning the front is unusually expensive relative to the back, the setup is on.
• The structure: Sell the near-dated at-the-money option, buy the far-dated option at the same strike. That is a long calendar spread.
• Two engines of profit: First, if the slope normalises, the front leg cheapens relative to the back and the spread widens in your favour. Second, and often larger, the front option decays faster than the back option. Time decay accelerates as expiry approaches, so the thing you are short is bleeding faster than the thing you are long. You are net collecting theta.
• The hedge: The two legs have different deltas and different gammas. You must delta hedge the combined position or you are simply taking a directional bet you did not intend.

Parameters (knobs)

• Expiry pair: One month against three months is a common choice. A wider gap means more decay differential but also more vega exposure on the back leg.
• Slope entry threshold: Only enter when the slope is in the top decile or quartile of its own history. Running this always on means paying to be in the trade during the calm periods when there is nothing to revert.
• Strike: At the money maximises theta and vega, which is what the trade is about. Moving the strike away from the money turns it into a directional bet on where spot will land.
• Exit rule: Hold to front expiry, or exit when the slope crosses back below its median. Slope-based exits are usually the better behaved of the two.

Where it works and where it hurts

It works after a panic that turns out to be an overreaction. The market drops 5 percent on some headline, front-month implied vol jumps from 15 to 30, the back month barely moves, and the curve inverts sharply. If the headline turns out to be noise, the front-month vol collapses within days, the short leg you sold decays fast, and the spread pays.

It hurts in two ways, and they are quite different:

• The gamma problem. The short front-month leg is where nearly all of the gamma lives. If the underlying makes a large move before that front option expires, the short leg loses money fast, and it loses far more than the modest cost of the spread would lead you to expect. You put on a trade that cost a small net debit and you can lose several multiples of it.
• The vega problem. You are long vega on the back leg. If overall volatility collapses across the whole curve, the back-month option you own loses value, and that can happen even while the slope moves in your favour. Being right on the slope and wrong on the level is a real and common way to lose here.

And of course the inversion may simply be correct. Front-month vol is elevated because something genuinely bad is unfolding. Selling it because it looks statistically expensive is, in that case, selling the crash insurance right before the crash.

Backtest design checklist

• Real surface data is mandatory. The trade is a bet on the relationship between two points on the implied vol curve. Model prices cannot represent that.
• Four legs of transaction cost. Buy and sell on each of two expiries, plus the delta hedging on top. Calendar spreads look profitable on paper and much less so after costs, which is one reason they are more popular in books than in production.
• Include the delta hedging P&L, and test the hedge frequency, because it materially changes the outcome.
• Attribute the P&L into slope change, level change and theta. If you find the profits came almost entirely from vega level rather than from the slope reverting, you have a volatility bet, not a term structure bet, and you should say so.
• Test the regime where inversion persists. Backtest through prolonged crises where the curve stayed inverted for weeks. That is the scenario the naive mean-reversion logic assumes away.

Common failure modes

• Underestimating short-leg gamma. The maximum loss on a calendar spread is not the net debit. That is a very common and very costly misconception.
• Trading the slope without hedging the level. You end up long vega without meaning to, and a vol crush punishes you for being right.
• Selling an inversion that deserves to be there. Statistical richness is not the same as mispricing. Sometimes the market knows something.
• Ignoring earnings and events. A single-name calendar spread that straddles an earnings date has a completely different risk profile, because the front-month vol is elevated for an entirely rational reason.

Our notes and suggestions

Calendar spreads are one of the more genuinely interesting volatility trades because they are not purely a short-volatility bet. You are long an option and short an option, so you have some natural offset in a vol spike, and the net vega position can even be arranged to be positive. That makes the tail less horrifying than in naked put writing.

But do not let that lull you. The short front leg's gamma is a real, sharp risk, and the transaction costs on four legs plus hedging are a serious headwind. This is a trade that needs institutional-quality execution to be worth doing at all.

Run it as an event-driven trade rather than a continuous one: wait for a genuine inversion, take the trade, hedge the delta, exit on normalisation. What would change our mind: term structure inversions becoming persistent rather than mean reverting, which would mean the market has repriced short-dated risk permanently and there is no reversion left to harvest.