Study 10: The Only Calendar Spread Strategy That Actually Makes Money
We tested calendar spreads on every S&P 500 stock for 20 years. They lost money. Then we figured out why — and built a version that works. Sharpe 0.677, win rate 56%, deployed only 17% of the time.
The Setup
Study 9 proved that calendar spreads are the best options structure across all market regimes — highest risk-adjusted return, positive Sharpe in every VIX bucket, strongest performance in recessions.
The obvious next question: can you build a systematic strategy around that?
We tried. The first version scored 400+ individual S&P 500 stocks each month on four factors — HV ratio, IV rank, price stability, trend neutrality — took the top 20, and ran calendar spreads on them. The result was a Sharpe of -0.785. Worse than not trading at all.
So we went back and figured out exactly why.
Why Individual Stocks Fail
The problem isn’t the scoring. It’s the math.
A calendar spread makes money when the underlying pins near the strike at front-month expiry. The profitable window is roughly ±4% around ATM. Outside that, the front leg either expires deep ITM (you owe on the short call) or the back leg has decayed too far to recover.
The expected 30-day move for an individual stock with 27% HV30 is:
0.27 × √(30/252) × S₀ ≈ 9.3% of stock priceThat’s more than twice the calendar’s break-even range. Individual stocks move too far, too fast. No amount of pre-trade scoring can fix this — if a stock’s expected move exceeds the strategy’s profitable range, the edge is structurally negative.
This is why the win rate on individual stock calendars sat at 33% regardless of score. The scoring was filtering noise on top of a losing P&L profile.
The original regime study worked because it was run on SPY and QQQ — indices with 15% HV30, expected 30-day move ≈ 5%. That’s inside the calendar’s profitable window.
The fix is simple: stop trading individual stocks, trade the indices.
The Strategy
Universe: 9 liquid ETFs with no earnings binary risk
- SPY, QQQ, IWM, DIA (broad market)
- TLT (20-yr Treasuries)
- XLV, XLE, XLF (sector)
- GLD (macro hedge)
Regime gate: Only enter when VIX regime is HIGH (15–35) or CRISIS (≥35). Sit on cash in LOW and MID VIX environments.
Execution: Each active month, rank ETFs by HV10/HV30 ratio. Trade top 6. Sell 30-DTE ATM call, buy 60-DTE ATM call at same strike. Exit at front-month expiry; mark back-month to model.
Why this gate? LOW and MID VIX means thin premiums and strong directional bias. The net debit on a calendar is smaller, the break-even range is tighter relative to typical moves, and selection noise dominates. HIGH/CRISIS VIX means premium is fat, term structure is steep, and mean-reversion pressure is elevated — exactly the conditions where calendar spreads thrive.
What we excluded: XLK (tech) and XLU (utilities) were removed after backtesting showed they consistently underperformed. Tech has large, asymmetric moves even in high-VIX environments. Utilities behave unexpectedly in rate-driven selloffs. Every ETF remaining in the universe has a positive Sharpe over the 20-year test period.
Results: 20 Years, 264 Trades
The backtest covers 2005–2025: the 2008–09 financial crisis, the 2011 European debt crisis, COVID-19, the 2022 rate shock. Every major volatility event included.
Overall performance:
| Metric | Value |
|---|---|
| Sharpe ratio | 0.677 |
| Win rate | 56.4% |
| Mean return per trade | +13.3% on capital deployed |
| Total net P&L | $4,596 per contract |
| Active months | 44 of 252 (17%) |
The strategy deploys capital in 44 months over 20 years. The other 206 months — 82% of the time — it does nothing. Cash earns T-bill rate. The compounding of those idle months matters: you’re not taking losses in unfavorable regimes, and you’re not tying up margin.
By VIX regime:
| VIX Bucket | Sharpe | Trades |
|---|---|---|
| CRISIS (≥35) | +1.155 | 66 |
| HIGH (15–35) | +0.513 | 198 |
CRISIS is the strongest regime by a wide margin. When VIX spikes above 35, the term structure is steep, near-term premium is elevated, and mean-reversion pressure is high. Calendar spreads are capturing exactly that. The HIGH regime is profitable but more modest — premiums are elevated but not extreme, and some months still have directional bias.
By market trend:
| Trend | Sharpe | Trades |
|---|---|---|
| BEAR | +0.919 | 138 |
| SIDEWAYS | +1.014 | 60 |
| BULL | −0.058 | 66 |
This is the cleanest signal in the data. Calendars are structurally neutral — they need the underlying to stay near the strike. In bear and sideways markets that condition holds. In bull markets the underlying drifts away from ATM, and the strategy barely breaks even.
Practically: if VIX is HIGH and the market is in a confirmed bull trend, be more conservative. The CRISIS filter largely handles this (CRISIS + BULL is rare), but it’s worth noting.
By ETF:
| ETF | Sharpe | Mean Return |
|---|---|---|
| TLT | +1.104 | +23.4% |
| XLV | +1.083 | +21.2% |
| XLE | +0.955 | +19.9% |
| IWM | +0.897 | +19.2% |
| QQQ | +0.457 | +9.4% |
| SPY | +0.418 | +8.0% |
| XLF | +0.466 | +8.7% |
| GLD | +0.318 | +5.3% |
| DIA | +0.371 | +7.5% |
TLT and XLV lead by a significant margin. Both make intuitive sense. Bonds spike in risk-off environments then mean-revert — the HV10/HV30 ratio elevated during CRISIS regimes, and TLT returns to fair value predictably. Healthcare is defensive: it moves less than the broad market in selloffs, pins near ATM more often.
Energy (XLE) is more volatile but moves in bursts around commodity events that are mean-reverting at 30-day horizons. IWM (small-caps) has elevated vol in stress but also strong reversion.
Recession vs expansion:
| Period | Mean Return |
|---|---|
| Recession | +25.8% |
| Expansion | +8.1% |
Recessions are when this strategy earns most of its money. VIX is structurally high, directional bias is less defined, and the calendar’s neutral P&L profile is an advantage rather than a liability.
The Edge in Plain Terms
Three things create the edge, in order of importance:
1. Regime timing. Not trading during 83% of months eliminates the losing trades. The largest single driver of returns is the months you skip, not the months you enter. LOW VIX calendar spreads are not just lower return — they have negative expected value. The VIX gate is the strategy.
2. ETFs over individual stocks. Removing earnings binary risk and idiosyncratic volatility means the underlying’s expected 30-day move is inside the calendar’s profitable range. This alone explains the difference between -0.785 Sharpe on individual stocks and +0.677 on ETFs.
3. Asset selection. Trading TLT, XLV, XLE, and IWM in high-vol regimes targets the instruments most likely to mean-revert after a spike. These are not stable, low-vol instruments — they’re volatile instruments with strong reversion tendencies, which is exactly what a calendar spread needs.
Practical Implementation
Position sizing: Each calendar spread risks only the net debit paid (approximately 1–3% of notional). Treating each spread as a fixed-dollar risk unit and running 6 ETFs per month produces consistent exposure.
Entry signal: First trading day of each month, check the VIX regime. If HIGH or CRISIS, pull the 9 ETFs, rank by HV10/HV30 ratio, sell the top 6. Use the next monthly expiry as front leg, the following month as back leg.
Exit: At front-month expiry, close both legs. The back-month mark-to-model value at expiry captures the remaining time premium. In practice, close the back leg immediately on expiry day to avoid pin risk.
What to watch: The VIX regime can shift mid-month. If VIX drops from HIGH to LOW within a cycle, the position is already on — hold it to expiry rather than exiting early. The edge is in the entry filter, not intramonth management.
What This Is Not
This is not a high-frequency strategy. 264 trades over 20 years is roughly one active cycle per month when markets are stressed, nothing otherwise. It does not compound quickly. It does not work in bull markets.
What it is: a defined-risk, positive-expectancy structure that activates exactly when most traders are scared and reactive. Calendar spreads require patience — the payoff is collected over the holding period, not at entry. The regime gate ensures you’re only entering when the underlying conditions support the strategy’s mechanics.
The math works because you’re not fighting the market. You’re waiting for the market to create the conditions you need, then stepping in.
Backtest methodology: 20-year synthetic options pricing via Black-Scholes, HV30 × 1.20 as IV proxy, 30/60 DTE calendar spreads on 9 liquid ETFs, regime classification via VIX, SPX 200-day MA, and NBER recession indicator. Commission $1.30 per contract leg.