Options Backtest

Selling Earnings Volatility: Short Straddles Across the Magnificent Seven

Options markets systematically overprice earnings moves. This study tests whether that overpricing is large enough and consistent enough to profit from, using short ATM straddles on the Magnificent Seven across Q1 2019 through Q3 2025.

AuthorBrian Liew, BSc Accounting and Finance, LSE
PublishedJanuary 2026
PeriodQ1 2019 to Q3 2025, n=189 events
DataOptionMetrics and IBES via WRDS
GitHub Code
Earnings Events
189
Q1 2019 to Q3 2025, 7 tickers
NVDA Win Rate
74.1%
Highest win rate in the study
AAPL Sharpe
1.93
Best risk-adjusted return
Combined P&L
$682
All 7 tickers, 189 events

The Strategy: Selling the Earnings IV Premium

In the days leading up to earnings, implied volatility rises as market participants pay up for optionality. Once earnings are released, the uncertainty resolves and IV collapses, regardless of whether the stock moves up or down. A short straddle profits when the actual stock move is smaller than the move priced into the straddle at entry. That gap is the IV crush.

Entry
T-0 Close
30 min before close on earnings day
Exit
T+1 Afternoon
Following trading day, after IV crush
P&L Identity
IV − RV
Premium received minus absolute stock move

The universe is restricted to the Magnificent Seven (AAPL, MSFT, NVDA, AMZN, GOOGL, META, TSLA) for three reasons: deep options liquidity, well-defined quarterly earnings calendars, and sufficient OptionMetrics history to produce 27 events per ticker over the study window.

Margin assumption. Margin is estimated at 20% of notional (spot price × 100 shares per contract), consistent with standard broker requirements for naked short options on equity underlyings.

Data Sources and P&L Construction

Options data was sourced from OptionMetrics IvyDB via WRDS (optionm.opprcd{year}). End-of-day closing bid and ask quotes were used to compute the mid-price straddle cost, serving as a proxy for the 3:45–4:00pm entry price. Earnings announcement dates were obtained from the IBES Actuals database (ibes.act_epsus), with ticker corrections applied for Alphabet (GOOG) and Meta (FB).

DimensionDetail
UniverseAAPL, MSFT, NVDA, AMZN, GOOGL, META, TSLA
PeriodQ1 2019 to Q3 2025 (27 events per ticker, 189 total)
InstrumentATM short straddle on nearest weekly expiry covering the earnings date
EntryApproximately 30 minutes before market close on earnings day (end-of-day OM quote used as proxy)
ExitFollowing afternoon, valued at intrinsic value (absolute stock move)
P&LGross, excluding bid-ask spread and commissions. Slightly overstates profitability on wins
Margin20% of notional (spot × 100) per contract, per standard broker requirements
DataOptionMetrics IvyDB and IBES Actuals via WRDS
Exit approximation. Valuing the exit at intrinsic value slightly overstates profitability on winning trades, since the straddle retains some time value and crossing the bid-ask spread on exit reduces realised P&L further. Results should be treated as directionally accurate rather than precisely replicable.

Summary Results by Ticker

Across 189 earnings events from Q1 2019 through Q3 2025, the strategy produced a combined total P&L of $682 trading one contract per event, essentially break-even. The aggregate result masks extreme cross-sectional dispersion: four of seven stocks generated positive total P&L, while AMZN, GOOGL, and META were net destroyers of capital, with AMZN alone accounting for a $22,304 loss driven by a catastrophic 2022 drawdown.

AAPL
Moderate
+$3,884
Sharpe 1.93
MSFT
Moderate
+$3,436
Sharpe 1.46
NVDA
★ Strong
+$5,340
Win rate 74.1%
AMZN
Avoid
−$22,304
Calmar −0.07
GOOGL
Avoid
−$4,583
Sharpe 0.05
META
Avoid
−$6,918
Sharpe −1.17
TSLA
Selective
+$21,827
Regime-dependent
Figure 1: Total P&L per Ticker (1 contract per earnings event)
Figure 2: Average Implied Move vs Average Realised Move (%)

Stocks where implied exceeds realised exhibit systematic IV overpricing, the precondition for a profitable short straddle. META is the only stock where realised moves consistently exceeded implied.

Figure 3: Risk-Adjusted Return Metrics per Ticker
Figure 4: Win Rate (%) vs Payoff Ratio

No stock combines a win rate above 70% with a payoff ratio above 1.0. NVDA leads on win rate (74.1%) but its payoff ratio of 0.48x means each loss averages 2x the average win. TSLA has the highest payoff (2.17x) but only a 51.9% win rate, driven by outsized winners in 2020–2021.

Headline finding. This strategy is stock-specific, not a general edge. TSLA accounts for $21,827 of the combined total, masking losses elsewhere. Remove TSLA and AMZN and the remaining five tickers sum to +$1,159 total over seven years, roughly $166 per ticker per year.

Year-by-Year Breakdown: Regime Dependency

The strategy’s profitability is demonstrably non-stationary. Results vary dramatically by year and by ticker, driven by shifts in realised volatility, macro regimes, and company-specific fundamentals. Select a ticker below to view its annual P&L.

Annual P&L
Key regime observations.

2020 (COVID volatility): Mixed results. TSLA surged to +$19,822. AMZN lost $9,194 on violent moves. MSFT and NVDA remained profitable.

2022 (rate shock): Most destructive year in the study. AMZN lost $36,920. META lost $7,673. NVDA was the exception at +$2,358.

2021: Best year broadly. Low realised vol and elevated premiums. TSLA generated +$8,339.

2023–2024: TSLA deteriorated to 0% win rate in both years (combined −$7,381). MSFT and AAPL remained constructive.

2025: Partial recovery across most tickers. AAPL (+$1,451), TSLA (+$2,670), META (+$1,472). MSFT was the notable miss at −$2,411.

Comprehensive Risk Metrics by Ticker

Ticker Win% Avg Edge% Payoff Sharpe Sortino Calmar Max DD Skew Kurt ROM%
AAPL 66.7% 0.99% 0.970 1.925 2.237 0.217 −$2,655 −2.848 9.926 4.30%
MSFT 63.0% 1.05% 0.941 1.463 2.730 0.173 −$2,948 −0.078 −0.394 3.05%
NVDA 74.1% 1.79% 0.476 1.309 1.152 0.111 −$7,130 −1.498 2.723 6.60%
AMZN 59.3% 0.19% 0.466 0.200 0.316 −0.068 −$48,819 −1.253 2.065 0.65%
GOOGL 51.9% 0.37% 0.801 0.050 0.098 −0.040 −$17,078 0.008 0.644 0.15%
META 55.6% −1.34% 0.566 −1.172 −1.263 −0.071 −$14,434 −1.079 1.348 −7.44%
TSLA 51.9% 0.50% 2.170 −0.079 −0.109 0.327 −$9,882 1.945 5.153 −0.41%

ROM = Return on Margin per trade. Sharpe, Sortino, Calmar computed on a per-trade basis. Skewness below −1 and kurtosis above 3 indicate meaningful tail risk.

Stock-by-Stock Assessment

NVDA★ Strong Edge: Highest Win Rate in the Study

NVDA leads on win rate (74.1%), avg edge (1.79%), and return on margin (6.60%), the strongest combination of any ticker in this study. The Sharpe of 1.31 reflects a genuine, persistent edge. The caveat is the payoff ratio of 0.48x: each losing trade is on average 2x the size of a winning trade, so the win rate must stay high for the strategy to remain profitable. The 2024 drawdown (−$4,314) and kurtosis of 2.72 reflect this tail risk. NVDA is best deployed when current implied moves appear elevated relative to historical realisation rates.

AAPLPositive Edge: Best Sharpe in the Study

AAPL delivers the highest Sharpe ratio in the study (1.93) with the shallowest max drawdown (−$2,655). A 66.7% win rate and avg edge of 0.99% are modest but consistent across all seven years. The 2025 result (+$1,451) confirms the pattern has held into the extended sample. The critical warning: skewness of −2.85 and kurtosis of 9.93 are the most extreme in the study. Rare but severe outlier losses can negate multiple years of gains. Strict position sizing is required.

MSFTPositive Edge: Reliable but Edge Has Compressed

MSFT still shows a real edge: Sharpe 1.46, Sortino 2.73, and the second-shallowest max drawdown in the study (−$2,948, behind AAPL’s −$2,655). However, the 2025 data changed the story materially. Win rate dropped from a prior 75% to 63.0%, and 2025 was MSFT’s first losing year in this study (−$2,411). Total P&L of $3,436 over seven years is thin relative to the capital commitment. The old characterisation of MSFT as the single best expression of this strategy no longer holds. It remains a reasonable trade with tight risk control, but not the dominant pick.

TSLAPositive Total P&L: Highly Regime-Dependent

TSLA accounts for $21,827 of the combined $682 total, making it the single driver of the study’s headline result. A payoff ratio of 2.17x with a 51.9% win rate produced a net profit, but the distribution is extreme: nearly all gains came from 2020–2021. From 2022 through 2024, TSLA went 6 for 16 (37.5% win rate) and lost $7,689 across those three years. 2025 recovered slightly (+$2,670). A trader running this strategy on TSLA since 2022 would be net negative. The edge existed in a specific macro regime and has not consistently re-emerged.

GOOGLAvoid: Negative Outcome Across Full Sample

GOOGL produced a total loss of −$4,583 across 27 events (Sharpe 0.05, Calmar −0.04). The market prices GOOGL earnings moves fairly on average: implied (5.52%) and realised (5.14%) are close, leaving minimal edge. Large single-year losses in 2019 (−$4,317) and 2021 (−$5,778) dominate the outcome. A win rate of 51.9% with a payoff ratio of 0.80x confirms the structural difficulty. There is no statistical basis for running this strategy on GOOGL.

METAAvoid: Negative Edge

META is a clear avoid. The avg realised move (8.90%) exceeded the avg implied move (7.56%), meaning options systematically underpriced earnings moves. The avg edge is −1.34%, second worst in the sample. The 2022 destruction (−$7,673 in one year) reflects META’s extraordinary fundamental repricing. Sharpe of −1.17 and ROM of −7.44% per trade are among the worst in the study. 2025 showed a partial recovery (+$1,472) but not enough to offset the structural negative edge.

AMZNAvoid: Catastrophic Tail Risk

AMZN is the most structurally dangerous ticker in this study. Despite a 59.3% win rate, AMZN produced a total loss of −$22,304 driven by 2022 (−$36,920 in a single year). The payoff ratio of 0.47x means losses are on average 2x larger than wins. The max drawdown of −$48,819 would require roughly 15 average winning trades to recover. 2019 (+$12,066) and 2021 (+$9,747) were large positive years, but the 2022 destruction wiped out both and then some. Do not run this strategy on AMZN.

Strategy-Level Risk Considerations

Tail risk and short vol asymmetry. Short straddles have theoretically unlimited loss potential. Five of seven tickers exhibit negative P&L skewness. AAPL’s kurtosis of 9.93 indicates an extremely fat-tailed distribution. A single severe earnings shock can negate years of small gains.
Transaction costs. All P&L figures use mid-prices. In practice, selling a straddle requires crossing the bid-ask spread on both legs at entry and exit. For a liquid name like AAPL this could be $40–$100 per contract round trip. The true edge, net of transaction costs, will be meaningfully lower than reported here.
Regime dependency. The strategy’s profitability is demonstrably non-stationary. 2022 was destructive across multiple tickers. TSLA’s edge has reversed since 2022. Any live deployment must monitor whether the fundamental conditions for IV overpricing still hold before each earnings event.
Margin and liquidity. Short naked straddles require significant margin. AMZN at $32,466 per contract ties up meaningful capital. In volatile markets, brokers can increase margin requirements intraday, potentially forcing unwanted position closure. Size positions so that margin represents no more than 2–5% of total account equity per trade.
EOD data approximation. This study uses end-of-day options prices as a proxy for entry at 3:45pm. IV may drift during the final 15 minutes. Results should be treated as directionally accurate rather than precisely replicable.

Trader Verdict: Stock-Specific, Not a General Edge

The combined P&L across 189 events and seven years is $682, essentially flat. That headline conceals wide dispersion: TSLA accounts for $21,827 of gains and AMZN accounts for $22,304 of losses. Remove those two outliers and the remaining five tickers sum to +$1,159, or roughly $166 per ticker per year. This is not a general edge; it is a narrow, stock-specific trade that requires careful selection.

Trade
NVDA · AAPL · MSFT
Positive edge, manageable risk, consistent positive Sharpe across the full sample
Selective
TSLA
High total P&L driven by 2020–2021 regime. Edge has deteriorated materially post-2022
Avoid
AMZN · META · GOOGL
Negative expected value or catastrophic tail risk. Do not run this strategy on these tickers

If forced to pick one ticker, AAPL is the most defensible choice: the highest Sharpe (1.93), the shallowest max drawdown in the study (−$2,655), and seven years of consistently positive aggregate results. The fat tail (kurtosis 9.93) demands that position sizing treat this as a rare-blowout strategy rather than a steady income stream. NVDA is the higher-conviction pick for traders comfortable with a 0.48 payoff ratio contingent on maintaining a 74% or higher win rate.

For live deployment: run pre-trade screens comparing the current ATM straddle price against the historical distribution of realised earnings moves for that specific ticker. Any event where the current implied move falls below the 40th percentile of historical realised moves should be skipped. Do not run the strategy indiscriminately across all seven tickers: the portfolio-level result is near zero and AMZN and GOOGL drag materially.

Disclaimer. Past performance does not guarantee future results. This research is for informational purposes only and does not constitute investment advice. Short options positions carry significant risk of loss, including loss exceeding initial margin.